|Home | About | Journals | Submit | Contact Us | Français|
The mammalian intestine is colonized with a diverse community of bacteria that perform many beneficial functions but can threaten host health upon tissue invasion. Epithelial cell-intrinsic innate immune responses are essential to limit the invasion of both commensal and pathogenic bacteria and maintain beneficial host-bacterial relationships; however, little is known about the role of various cellular processes, notably autophagy, in controlling bacterial interactions with the intestinal epithelium in vivo. We demonstrate that intestinal epithelial cell autophagy protects against tissue invasion by both opportunistically invasive commensals and the invasive intestinal pathogen, Salmonella typhimurium. Autophagy is activated following bacterial invasion of epithelial cells through a process requiring epithelial cell-intrinsic signaling via the innate immune adaptor protein MyD88. Additionally, mice deficient in intestinal epithelial cell autophagy exhibit increased dissemination of invasive bacteria to extraintestinal sites. Thus, autophagy is an important epithelial cell-autonomous mechanism of antibacterial defense that protects against dissemination of intestinal bacteria.
The mammalian intestine is home to a complex and diverse population of bacteria. The members of this microbial community are essential for host metabolism and digestion, thereby creating a mutualistic relationship. Although the overall host-microbial relationship is symbiotic, the bacterial components of this community span a wide spectrum of lifestyles, encompassing benign commensals, opportunistically pathogenic pathobionts, and overt pathogens.
The intestinal epithelium interfaces directly with this diverse community of bacteria, and is the first line of defense against bacterial invasion of host tissues. The epithelium must therefore be equipped with a diverse array of defenses against bacterial attachment and invasion. One mechanism by which the epithelium defends against bacterial penetration is by secreting antimicrobial proteins that limit bacterial contact with the epithelial surface (Vaishnava et al., 2011). However, certain intestinal pathogens, such as Salmonella enterica Serovar Typhimurium (Salmonella typhimurium), or opportunistically invasive commensal bacteria, such as Enterococcus faecalis, can evade this first line of innate defense and enter epithelial cells (Eichelberg and Galán, 1999; Klare et al., 2001; Müller et al., 2012). This raises the question of whether there are epithelial cell-intrinsic immune mechanisms that detect invading bacteria and limit their further dissemination.
Autophagy is an evolutionarily ancient process in which cytoplasmic materials are targeted to the lysosome for degradation. Portions of the cytoplasm are sequestered into double-membrane structures, called autophagosomes, which fuse with lysosomes, delivering their contents for degradation by lysosomal enzymes (Deretic and Levine, 2009). The process involves the concerted action of several cytoplasmic proteins. These include LC3, which becomes lipid-conjugated and associates with the autophagosome membrane, and Atg5, which is conjugated to Atg12 and associates with the elongating isolation membrane (Mizushima et al., 2002). A primary function of autophagy is to maintain cellular homeostasis by degrading cytoplasmic contents during cellular starvation and by recycling damaged organelles and proteins (Rabinowitz and White, 2010).
More recently, autophagy has been shown to be critical for the recognition and degradation of intracellular pathogens, thus functioning as an innate barrier to infection (Deretic and Levine, 2009; Levine et al., 2011). Bacterial targets of autophagy include the intestinal pathogens S. typhimurium (Jia et al., 2009; Kuballa et al., 2008; Rioux et al., 2007; Wild et al., 2011) and Listeria monocytogenes (Py et al., 2007). Autophagy limits the replication of both of these bacterial species in cell culture models (Py et al., 2007; Wild et al., 2011). Intestinal epithelial cell autophagy is also critical for resistance against S. typhimurium infection in the nematode Caenhorhabditis elegans (Jia et al., 2009). However, the importance of autophagy for mammalian intestinal immunity remains underexplored. Such functions are likely to be especially important in the intestinal epithelium, which interfaces with a dense microbial community that harbors invasive bacteria.
Genetic studies of inflammatory bowel disease (IBD) have revealed important roles for autophagy pathway proteins in intestinal immune homeostasis. IBD is a chronic inflammatory disease of the intestine that arises from dysregulated interactions with the resident microbiota (Xavier and Podolsky, 2007). Recent studies have identified polymorphisms in genes of the autophagy pathway that are linked to Crohn’s disease, a type of IBD in which the inflammation is localized to the distal small intestine and variable regions in the colon. Mutations in the critical autophagy gene ATG16L1 are associated with a predisposition to Crohn’s disease in humans (Hampe et al., 2007; Rioux et al., 2007; Wellcome Trust Case Control Consortium, 2007). However, the Atg16L1 polymorphisms associated with Crohn’s disease do not confer autophagy defects in mice, suggesting that the inflammatory phenotypes arise from autophagy-independent functions of Atg16L1 (Cadwell et al., 2008). Rather, the mutations cause defects in granule formation in Paneth cells, a specialized epithelial cell lineage that secretes abundant antimicrobial proteins (Cadwell et al., 2008; 2010). Thus, it is not yet clear whether bona fide autophagy plays a role in maintaining intestinal homeostasis in vivo.
Here we report that intestinal epithelial cell autophagy is essential for mammalian intestinal defense against invasive bacteria. We show that epithelial autophagy is activated in the mouse intestinal epithelium by a pathogen, S. typhimurium, as well as by E. faecalis, an opportunistically invasive commensal. Further, we find that epithelial autophagy is specifically triggered by bacterial invasion of epithelial cells and requires epithelial cell-intrinsic MyD88 signaling. Finally, we use mice with an epithelial cell-specific deletion of a critical autophagy factor to show that epithelial cell autophagy is critical for limiting extraintestinal spread of S. typhimurium. Our observations thus reveal that autophagy is a key epithelial cell-autonomous mechanism of antibacterial defense that protects against dissemination of intestinal bacteria. Our findings provide insight into how the mammalian intestinal epithelium maintains homeostasis with a diverse intestinal microbiota and establish a key role for autophagy in innate immune defense of the intestine.
LC3 is an essential autophagy protein that is recruited from the cytoplasm to the autophagosome membrane (Mizushima et al., 2002). A classical assay for autophagy activation uses immunofluorescence to visually assess the recruitment of cytoplasmic LC3 to autophagosomes, which are detected as morphologically distinct punctate structures (Mizushima et al., 2010). To determine whether bacteria activate autophagy in the intestinal epithelium in vivo, we assessed LC3 distribution in the intestinal epithelial cells of mice of differing bacterial colonization status. Germ-free mice are microbiologically sterile, thus allowing us to compare LC3 localization in the presence and absence of intestinal bacteria. Both germ-free and conventionally-raised (conventional) mice exhibited a diffuse, cytoplasmic distribution of LC3, indicating that a specified pathogen-free (SPF) microbiota was not sufficient to activate epithelial autophagosome formation as indicated by LC3 staining (Figure 1A).
To test whether an invasive bacterial pathogen could elicit autophagosome formation we orally infected mice with S. typhimurium. 24 hours after S. typhimurium inoculation into both germ-free and conventional mice, we observed punctate LC3+ structures in small intestinal epithelial cells, indicating autophagosome formation (Figure 1A). The LC3+ puncta were positioned apically relative to the nucleus following S. typhimurium infection of germ-free mice, whereas they were positioned both apically and basolaterally relative to the epithelial cell nuclei in conventional mice (Figure 1A). Although the cause of this difference is not clear, it may be due to different routes of S. typhimurium entry (e.g., apical versus basolateral) in the two different host settings (Chieppa et al., 2006; Müller et al., 2012; Niess, 2005). We also performed Z-stack reconstructions of the fluorescent images in multiple focal planes to verify that the LC3+ structures were located within epithelial cells (Movie S1).
We next characterized the location and timing of epithelial LC3+ autophagosome formation following S. typhimurium infection. Numbers of LC3+ puncta were highest in the distal small intestine (ileum) and diminished in the middle and proximal regions (jejunum and duodenum, respectively)(Figure 1B; Figure S1A). LC3+ autophagosomes were more abundant in epithelial cells inhabiting the ileal villus tips compared to the cells located closer to the crypts (Figure S1A). LC3+ puncta were also observed in colonic epithelial cells following S. typhimurium infection but were rare relative to the numbers in the terminal ileum (Figure S1B). Numbers of LC3+ autophagosomes were highest in the terminal ileum at ~24 hours following S. typhimurium infection, and diminished at 48 and 72 hours post-infection (Figure 1C). Thus, autophagosome formation is a rapid and transient response of the intestinal epithelium to oral S. typhimurium infection. Drawing on these initial observations, all subsequent analysis was performed on ileal tissues, with epithelial cells visualized at the midpoint between the crypt base and villus tip.
During recruitment to the autophagosome, LC3-I is lipidated to yield LC3-II, which becomes associated with the autophagosome membrane (Pankiv et al., 2007). Western blot analysis of isolated ileal epithelial cells showed increased conversion of LC3-I to LC3-II at 24 hours after S. typhimurium infection, consistent with autophagy activation (Figure 1D,E). This conversion was diminished after 72 hours (Figure 1D,E), which accords with the reduced numbers of LC3+ autophagosomes observed by immunofluorescence (Figure 1C).
To assess whether autophagosomes colocalized with intracellular bacteria, we orally challenged germ-free mice with S. typhimurium constitutively expressing green fluorescent protein (S. typhimurium-GFP). 24 hours after challenge, we could visualize S. typhimurium within enterocytes (Figure 1F). Analysis of serially-cut sections with a no primary antibody control verified that the GFP signal was not due to nonspecific autofluorescence (Figure S1C), which is frequently observed in fixed small intestinal tissues (Salzman et al., 2010; Vaishnava et al., 2011). The GFP signal was specific to S. typhimurium, as no signal was detected above background in uninfected germ-free tissues (Figure S1D). Upon merging the LC3 and GFP channels, we found that some S. typhimurium-GFP colocalized with LC3 (Figure 1F), consistent with the targeting of S. typhimurium to autophagosomes. The colocalization of LC3 and GFP was determined to have a Pearson’s coefficient of 0.89, indicating a strong, positive relationship between the two signals (Figure 1G). Approximately 40% of intracellular S. typhimurium colocalized with LC3 (Figure 1H), while ~70% of the LC3+ puncta colocalized with S. typhimurium-GFP (Figure 1I). These results demonstrate that autophagosome formation occurs in epithelial cells in response to S. typhimurium infection, and that the majority of autophagosomes colocalize with bacteria.
To determine whether the epithelial autophagic response to S. typhimurium was representative of a general response to invading bacteria, we asked whether epithelial autophagy could also be activated by opportunistically invasive members of the intestinal microbiota. To test this idea we orally challenged germ-free mice with Enterococcus faecalis, a Gram-positive, opportunistically invasive member of the human intestinal microbiota (Klare et al., 2001). 24 hours after challenge, we detected E. faecalis (harboring an episomal GFP-expressing plasmid) in epithelial cells of the small intestine (Figure 2A,D). Coincident with the ability of E. faecalis to invade epithelial cells, we observed the formation of numerous LC3+ autophagosomes (Figure 2B,E). In contrast to the apical localization of the LC3+ puncta observed after S. typhimurium colonization of germ-free mice (Figure 1A), the majority of the E. faecalis-induced autophagosomes were located on the basolateral side of the nucleus. As discussed above, we suggest that this may be due to differing routes of epithelial cell entry for S. typhimurium and E. faecalis.
Western blot analysis of isolated ileal epithelial cells revealed increased conversion of LC3-I to LC3-II at 24 hours following E. faecalis colonization, consistent with autophagy activation (Figure 2F,G). In contrast, a non-invasive member of the microbiota, Lactobacillus salivarius, colonized the small intestine to approximately the same extent as E. faecalis (Figure 2C) but was not detected within epithelial cells (Figure 2A,D) and did not trigger formation of LC3+ autophagosomes (Figure 2B,E). Together, these results suggest that epithelial cell autophagy can be activated by opportunistically invasive members of the microbiota.
We further characterized the autophagic response of intestinal epithelial cells using transmission electron microscopy. 24 hours after introduction of S. typhimurium into germ-free mice we observed double-membraned autophagosomes within small intestinal epithelial cells (Figure 3B,C)(Mizushima et al., 2010). The double membranes enclosed bacteria, and were absent from the intestinal epithelial cells of germ-free mice (Figure 3A). We also observed double-membrane structures surrounding bacteria in the epithelial cells of mice colonized for 24 hours with E. faecalis (Figure 3D,E). These results support the idea that invading bacteria activate autophagy within intestinal epithelial cells, and that bacteria are targeted to the autophagosomes.
Our findings above suggested that bacterial invasion of epithelial cells is required to activate intestinal epithelial autophagy. To further test this idea we used genetically altered S. typhimurium. The SPI-1 pathogenicity island encompasses genes essential for S. typhimurium entry into epithelial cells, and a S. typhimurium mutant engineered to lack this island (ΔSPI-1) is defective in its ability to invade gut epithelia (Eichelberg and Galán, 1999). The wild-type and ΔSPI-1 strains colonized germ-free mice to equivalent levels after 24 hours (Figure S2) yet the numbers of LC3+ autophagosomes formed in response to the ΔSPI-1 mutant were dramatically reduced relative to the wild-type strain (Figure 4A,B). This suggests that cellular invasion is required for S. typhimurium to activate epithelial autophagy. To corroborate this finding we tested a second isogenic S. typhimurium mutant strain lacking a single component of the type III secretion apparatus, InvA. Deletion of invA also inhibits epithelial cell entry by S. typhimurium (Everest et al., 1999; Galán et al., 1992), and like the ΔSPI-1 strain, the ΔinvA mutant elicited reduced numbers of LC3+ autophagosomes in the small intestinal epithelium (Figure 4A,B). These results support the idea that bacterial entry into epithelial cells is a prerequisite for autophagy activation.
Prior studies have shown that bacterial induction of autophagy in macrophages requires activation of innate immune signaling pathways (Shi and Kehrl, 2008; Travassos et al., 2010). At the same time, other intestinal epithelial cell-intrinsic innate immune responses, such as expression of the antimicrobial protein RegIIIγ, are dependent on MyD88 (Brandl et al., 2007; Vaishnava et al., 2011), which signals downstream of Toll-like receptors (TLRs), the IL-1 receptor (IL-1R), and the IL-18 receptor (IL-18R) (Akira et al., 2006). We therefore tested whether bacterial induction of epithelial autophagy was similarly dependent on MyD88. As in conventional wild-type mice, epithelial LC3 showed a cytoplasmic distribution in conventional Myd88-/- mice (Figure 5A,B). However, unlike wild-type mice, Myd88-/- mice did not show detectable LC3+ autophagosome formation after oral challenge with S. typhimurium (Figure 5A,B), indicating that MyD88 is essential for bacterial activation of autophagy in small intestinal epithelial cells.
We next investigated the cellular origin of the MyD88 signals required to elicit epithelial autophagy. We generated mice with an epithelial cell–specific deletion of Myd88 (Myd88ΔIEC) (Ismail et al., 2011) by crossing mice carrying a loxP-flanked (floxed, fl) Myd88 allele (Myd88fl/fl) (Hou et al., 2008) with mice expressing Cre recombinase under the control of the intestinal epithelial cell (IEC)–specific villin promoter (Madison et al., 2002). 24 hours after oral challenge with S. typhimurium, the Myd88ΔIEC mice retained a cytoplasmic distribution of epithelial LC3 whereas their Myd88fl/fl littermates showed LC3+ autophagosome formation (Figure 5C,D). Western blot analysis of isolated ileal epithelial cells revealed increased conversion of LC3-I to LC3-II at 24 hours following S. typhimurium colonization of the Myd88fl/fl mice but not the Myd88ΔIEC mice (Figure 5E,F), consistent with reduced autophagosome formation in the absence of epithelial MyD88. These results show that epithelial cell-intrinsic MyD88 signaling is required for bacterial induction of autophagy in intestinal epithelial cells.
We examined whether other innate immune signaling pathways are also required for intestinal epithelial autophagy. Prior studies have implicated signaling through the adaptor protein Toll-interleukin-1 receptor domain containing adaptor-inducing interferon-β (TRIF) in bacterially-activated autophagy in RAW macrophages (Xu et al., 2007). However, we observed autophagosome formation in TRIF-/- mice following oral S. typhimurium infection (Figure S3), indicating that TRIF is not required for intestinal epithelial autophagy in vivo.
Since the intracellular pattern recognition receptor Nod2 has been implicated in the autophagy pathway in macrophages (Travassos et al., 2010), dendritic cells (Cooney et al., 2010), and cultured epithelial cells (Homer et al., 2010), we also examined Nod2-/- mice (Kobayashi et al., 2005). We observed numerous LC3+ puncta in conventional Nod2-/- mice, even in the absence of an additional bacterial challenge (Figure 5G,H). Depletion of the microbiota with broad spectrum antibiotics reduced the number of LC3+ autophagosomes in the Nod2-/- mice (Figure 5G,H), indicating that autophagosome formation was a response to the microbiota and was probably not due to blocked flux through the autophagy pathway (Mizushima et al., 2010). Because epithelial autophagosome formation is associated with bacterial invasion into the cell cytoplasm (Figures 1--4),4), we postulated that the increased autophagosome formation in Nod2-/- mice was due to increased invasion of epithelial cells by commensal bacteria. To test this idea we assayed for intracellular bacteria using fluorescence in situ hybridization with a universal 16S ribosomal RNA (rRNA) gene probe. We observed increased numbers of bacteria within the small intestinal epithelial cells of Nod2-/- mice as compared to wild-type mice (Figure 5I,J). Numbers of intracellular bacteria were reduced upon antibiotic treatment of the Nod2-/- mice (Figure 5I,J). These findings support the idea that autophagy activation in the intestinal epithelium is associated with bacterial invasion of epithelial cells. Together, our results establish that in vivo intestinal epithelial autophagy is MyD88-dependent, TRIF-independent and NOD2-independent, and suggest that commensal bacteria can elicit epithelial autophagy in immune deficient hosts where there is increased bacterial invasion of the intestinal epithelium.
We next sought to determine whether autophagy in intestinal epithelial cells contributes to host resistance to bacterial infection and dissemination. We created mice with an intestinal epithelial cell-specific deletion of the essential autophagy gene Atg5 by crossing mice with a loxP-flanked Atg5 allele (Atg5fl/fl) (Hara et al., 2006; Tsukamoto et al., 2008) with villin-Cre transgenic mice (Madison et al., 2002) to produce Atg5ΔIEC mice. We verified that ATG5 expression was diminished in intestinal epithelial cells from the terminal ileum (Figure S4A). We further observed reduced numbers of LC3+ autophagosomes in the intestinal epithelial cells of the Atg5ΔIEC mice after oral S. typhimurium challenge as compared to Atg5fl/fl littermates (Figure 6A,B). The decreased numbers of autophagosomes accorded with reduced conversion of LC3-I to LC3-II in the Atg5ΔIEC mice (Figure 6C,D). While there was no histological evidence of overt inflammation or pathology in unchallenged Atg5ΔIEC mice (Figure S4B), the decreased autophagosome formation in the Atg5ΔIEC mice coincided with increased numbers of intracellular S. typhimurium 24 hours after oral challenge (Figure 6E,F). These data support the idea that intestinal epithelial autophagy is a cell autonomous mechanism that limits the persistence of invading bacteria.
The Atg5ΔIEC mice also showed increased S. typhimurium dissemination to extraintestinal tissues. 24 hours after oral S. typhimurium challenge, we recovered higher numbers of bacteria from the spleens and livers of the Atg5ΔIEC mice relative to their Atg5fl/fl littermates (Figure 6G,H). This was not due to differences in overall small intestinal bacterial numbers, which were similar between the two groups (Figure 6I). The differences in dissemination to extraintestinal tissues were only observed after oral infection; no significant difference was observed in the numbers of S. typhimurium recovered from the spleens of Atg5ΔIEC and Atg5fl/fl mice following intraperitoneal inoculation (Figure 6J). This indicates that the increased S. typhimurium dissemination was not due to a global immune defect in the Atg5ΔIEC mice.
We considered whether the increased dissemination of S. typhimurium in Atg5ΔIEC mice might be due to autophagy-independent effects of ATG5-deficiency (Zhao et al., 2008). The Atg5ΔIEC mice did not exhibit increased permeability to orally-administered FITC-dextran (Figure S4C), suggesting that the increase in S. typhimurium dissemination did not arise from enhanced nonspecific barrier permeability. We also did not observe increased dissemination of the non-invasive S. typhimurium strains ΔSPI-1 and ΔinvA in Atg5ΔIEC mice (Figure 6G and H). Thus, the increased dissemination of wild-type S. typhimurium in Atg5ΔIEC mice is unlikely to arise from defects in Paneth cell granule exocytosis and antimicrobial protein secretion that have been observed in these mice (Cadwell et al., 2008). Together, our findings show that intestinal epithelial autophagy plays an essential role in limiting the numbers of S. typhimurium within intestinal epithelial cells and in controlling its dissemination to extraintestinal tissues.
The intestinal epithelium plays diverse roles in immune protection against commensal and pathogenic bacteria in the intestine. For example, epithelial cells produce a variety of antimicrobial proteins that are critical for maintaining homeostasis with the microbiota (Salzman et al., 2010; Vaishnava et al., 2011). α-defensins produced by Paneth cells shape the composition of luminal bacterial communities in the small intestine (Salzman et al., 2010). Other antimicrobial proteins protect the apical surface of the epithelium from bacterial attachment. RegIIIγ, an antimicrobial lectin that is expressed throughout the intestinal epithelium (Cash et al., 2006), is regulated by bacterial signals through an epithelial cell-intrinsic MyD88 pathway (Brandl et al., 2007; Vaishnava et al., 2011). RegIIIγ interacts with the mucus layer to limit bacterial interactions with the intestinal epithelial surface and thus restricts bacterial invasion of host tissues (Vaishnava et al., 2011). In addition to secreting antimicrobial proteins, the intestinal epithelium also orchestrates adaptive immune responses to intestinal bacteria through the secretion of cytokines and other factors (Artis, 2008).
Despite the formidable antimicrobial barrier at the epithelial surface, some bacteria are equipped to evade these defenses and invade epithelial cells. Our findings suggest that autophagy is a critical mechanism of epithelial cell-intrinsic innate immunity that eliminates invading bacteria before they can access deeper tissues. Consistent with this idea, we found that epithelial autophagy is specifically activated by invasive bacteria. These include S. typhimurium, an overt pathogen, and E. faecalis, an opportunistically-invasive pathobiont. In contrast, non-invasive bacteria, such as Lactobacillus salivarius or S. typhimurium lacking the SPI-1 pathogenicity island, did not trigger autophagosome formation in the intestinal epithelium. This argues that autophagy is not activated as a general response to a SPF microbiota, but requires bacterial invasion into the cell cytoplasm.
The autophagy machinery is involved in cellular functions that are distinct from classical autophagy involving the targeting of bacteria to autophagosomes (Zhao et al., 2008). Mice harboring genetic mutations or deletions in Atg5 or Atg16L1 exhibit abnormalities in the secretory pathway of Paneth cells, a specialized small intestinal epithelial lineage involved in antimicrobial protein secretion (Cadwell et al., 2008; 2010). It is possible that the Atg5ΔIEC mice exhibit increased S. typhimurium invasion because of such defects in Paneth cell function, which could lead to increased accessibility of bacteria to the epithelial surface. However, a role for classical autophagy would be consistent with our finding that S. typhimurium challenge activates autophagosome formation in the intestinal epithelium. Additionally, we found that non-invasive S. typhimurium strains do not show increased dissemination in Atg5ΔIEC mice. This argues against the idea that the increased dissemination of wild-type S. typhimurium in Atg5ΔIEC mice is due to defective Paneth cell secretory function. Finally, it is possible that S. typhimurium infection elicits nutrient deprivation that initiates autophagy in order to sustain metabolic functions (Tattoli et al., 2012). However, our observation that S. typhimurium-induced autophagosome formation is also MyD88-dependent suggests that nutrient starvation does not entirely account for bacterially-induced autophagy in the intestinal epithelium.
Other intestinal bacterial pathogens may also be targets of epithelial cell autophagy. For example, epithelial cell expression of the autophagy protein Atg7 confers protection against the mouse intestinal pathogen Citrobacter rodentium (Inoue et al., 2012). C. rodentium is an attaching and effacing pathogen which forms lesions on the apical surface of the colonic epithelium but does not enter epithelial cells in large numbers (Mundy et al., 2005). It is not yet clear whether bona fide autophagy is involved in limiting C. rodentium intestinal colonization and pathogenesis (Inoue et al., 2012). Thus, it will be interesting to determine whether C. rodentium triggers autophagy in colonic epithelial cells, and whether autophagy-dependent or -independent functions of Atg7 are involved in protection against C. rodentium.
Our analysis of the host factors involved in autophagy initiation in the intestinal epithelium revealed a requirement for epithelial cell-intrinsic MyD88. However, the host receptors that lie upstream of MyD88 remain to be defined. Given that autophagy activation requires invasive bacteria, it is possible that intracellular TLRs, such as TLR9, might be involved. It is also possible that invading bacteria could penetrate the epithelium via a paracellular route, or could traverse and exit epithelial cells via the basolateral surface, and thus might trigger basolaterally oriented epithelial TLRs such as TLR5 (Gewirtz et al., 2001). Given the dependence of epithelial autophagosome formation on bacterial invasion, it seems unlikely that these responses are stimulated through activation of apically-oriented epithelial TLRs. It is also possible that autophagy activation could involve MyD88-dependent IL-1 or IL-18 signaling (Akira et al., 2006).
Another question is how MyD88 integrates with the autophagy pathway to promote autophagosome formation. One possibility is that MyD88 signaling controls the expression of proteins involved in autophagosome formation. Another possibility is that MyD88 acts as an adaptor for assembly of protein complexes that regulate autophagosome formation. In macrophages, MyD88 promotes autophagy by interacting with Beclin 1, a key autophagy protein that promotes autophagosome formation by recruiting other autophagy factors to the pre-autophagosomal membrane (Kihara et al., 2001a; 2001b). Beclin 1 function is inhibited by binding to Bcl-2, an autophagy inhibitor (Pattingre et al., 2005). The MyD88-Beclin 1 interaction inhibits the binding of Beclin 1 to Bcl-2, thus relieving the Bcl-2 inhibition and promoting autophagy (Shi and Kehrl, 2008). It will be interesting to determine whether MyD88 plays a similar role in the intestinal epithelium.
A growing body of research supports a connection between autophagy gene mutations and IBD. Genome wide association studies have linked mutations in Atg16L1 to an increased risk for developing Crohn’s Disease. The polymorphisms were found to impact autophagy-independent functions of Atg16L1, resulting in abnormal granule packaging and protein secretion in Paneth cells (Cadwell et al., 2008). These abnormalities were associated with increased susceptibility to Crohn’s disease-like pathologies upon virus infection of mice (Cadwell et al., 2010). These findings showed how defects in autophagy gene function can lead to inflammatory pathologies. Our results expand upon these findings by suggesting that autophagy defects could have a broader role in inflammatory disease by interfering with the ability of the intestinal epithelium to clear invading bacteria. This could lead to ongoing immune activation, particularly in the presence of other immune system defects or perturbed intestinal microbial communities.
C57BL/6 wild-type, Myd88-/-, Myd88fl/fl, Myd88ΔIEC, villin-Cre, and Myd88fl/fl mice were maintained in the barrier at the University of Texas Southwestern Medical Center. Germ-free C57BL/6 mice were maintained in isolators as described (Cash et al., 2006). Atg5fl/fl mice have been described previously (Hara et al., 2006). B6.SJL-Tg(Vil-cre)997Gum/J (villin-cre) mice were obtained from the Jackson Laboratory. Experimental mice were generated by mating Atg5fl/fl mice with Atg5fl/fl mice that were heterozygous for the villin-Cre transgene. Genotyping of the Atg5fl/fl and the Cre gene was described previously (Miller et al., 2008). 6-8 week old mice were used for all experiments. Experiments were performed using protocols approved by the Institutional Animal Care and Use Committees of the UT Southwestern Medical Center.
Salmonella enterica Serovar Typhimurium (SL1344) and its isogenic mutants ΔSPI-1 (Eichelberg and Galán, 1999), ΔinvA (Galán et al., 1992), and S. typhimurium-GFP (a gift from V. Sperandio, UT Southwestern Medical Center) were grown in Luria Broth at 37°C. Enterococcus faecalis (V583) and its derivatives E. faecalis (V583 pMV158gfp) (Nieto and Espinosa, 2003) were grown in BHI broth at 37°C. Lactobacillus salivarius (ATCC 11741) was transformed with the plasmid pMV158gfp to create the strain ATCC 11741 pMV158gfp.
Rabbit anti-ATG5 and anti-LC3 antibodies were from Novus Biologicals. Goat anti-GFP and donkey anti-goat IgG (conjugated to DyLight 488) were from Abcam. Alexa Fluor 568-Phalloidin (Invitrogen) was used to identify epithelial cell borders of OCT-embedded frozen small intestine sections fixed in 4% paraformaldehyde/10% sucrose.
Small intestines were fixed with Bouin’s fixative overnight at 4°C and embedded in paraffin. The tissues were washed twice for 10 min in xylene, twice in 100% EtOH, twice in 95% EtOH, and rinsed in deionized water for 5 min. Sections were boiled in 10 mM sodium citrate (pH 6), washed in PBS, blocked in 1% BSA, 0.3% Triton-X-100 in PBS, and incubated with rabbit anti-LC3 (1:200)(Novus Biologicals) at 4°C overnight. After washing in TBS-T (50 mM Tris, pH 7.6, 150 mM NaCl, 0.05% Tween-20), sections were incubated with goat anti-rabbit IgG-Cy3 (1:400 Biomeda) for 30 min at 25°C and counterstained with DAPI. Images were captured on a Zeiss AxioImager M1 microscope.
Ileums were flushed with 4°C PBS and epithelial cells were lysed in situ with Tissue–Protein Extraction Reagent (T-PER, Thermo Scientific) with complete protease inhibitor cocktail (Roche) for 5 min. Lysates were centrifuged at 4°C for 5 min, boiled, and analyzed by Western blotting.
Ileums were flushed with ice-cold PBS and fixed for 24 h at 4°C in 2% paraformaldehyde, 2% glutaraldehyde, 0.1 M cacodylate (in PBS), post-fixed for 1 h in 2% osmium tetroxide, and stained with a solution of aqueous uranyl acetate and lead. Samples were dehydrated in graded alcohols and embedded in Poly/BED 812 (Electron Microscopy Sciences). Sections were viewed with a TEM 2 JEOL 1200EX II or Scanning Electron microscope.
Mice were infected intraperitoneally with S. typhimurium at 103 CFU per mouse, or intragastrically by gavage with 109 CFU per mouse. CFU in the small intestine, liver, and spleen were determined by dilution plating on Luria broth plates containing 100 μg/ml ampicillin.
Mice were given ampicillin (1 mg/ml), vancomycin (500 μg/ml), neomycin (1 mg/ml), metronidazole (1 mg/ml) and streptomycin (1 mg/ml) in drinking water for 1 week. All antibiotics were from Sigma. Microbiota depletion was confirmed by aerobic and anaerobic culture of intestinal contents on Brain Heart Infusion agar containing 10% calf blood.
Small intestinal tissues were fixed in Carnoy’s fixative (Ricca Chemical), embedded in paraffin, and cut to 5 μm. Sections were hybridized to a universal bacterial probe directed against the 16S rRNA gene: 5’-Cy3-GCTGCCTCCCGTAGGAGT-Cy3-3’ as described (Vaishnava et al., 2011) and visualized using a Zeiss AxioImager M1 Microscope.
p values were calculated using the unpaired two-tailed Student’s t test. All experiments were performed at least three times and results are expressed as the mean±standard error of the mean (SEM).
We thank K. Ruhn, C.L. Behrendt, and C. Clements for assistance with mouse experiments, A. Orvedahl for advice on LC3 detection, L. Mueller for help with electron microscopy, N. Mizushima for Atg5fl/fl mice, and A. DeFranco for Myd88fl/fl mice. This work was supported by NIH R01 DK070855 (LVH), NIH U54AI057156 (BL), a Burroughs Wellcome Foundation Investigators in the Pathogenesis of Infectious Diseases Award (LVH), and the Howard Hughes Medical Institute (LVH, BL). JLB was supported by a UNCF/Merck Graduate Science Research Dissertation Fellowship.
Supplemental Information includes four Supplemental Figures, and one Supplemental Movie.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.