Volume 137, Issue 2, Pages 579-587.e2 (August 2009) Salmonella Induces Flagellin- and MyD88-Dependent Migration of Bacteria-Capturing Dendritic Cells Into the Gut Lumen  Juan L. Arques, Isabelle Hautefort, Kamal Ivory, Eugenio Bertelli, Marì Regoli, Simon Clare, Jay C.D. Hinton, Claudio Nicoletti  Gastroenterology  Volume 137, Issue 2, Pages 579-587.e2 (August 2009) DOI: 10.1053/j.gastro.2009.04.010 Copyright © 2009 AGA Institute Terms and Conditions

Figure 1 Confocal microscopic analysis of small intestinal mucosa of BALB/c mice after 3-hour challenge with PBS or GFP-expressing bacteria. Cryosections were immunostained for CD11c (red) and actin (blue) following challenge with PBS (A) and GFP-expressing E coli DH5α (B) and noninvasive S typhimurium SL1344 (ΔSPI1, rpsM::gfp) (C and D). Although both E coli (B, insets) and Salmonella (C and D, insets) were found inside the tissues, only Salmonella induced a marked recruitment of CD11c+ DCs. Salmonella occasionally induced the formation of an intraepithelial cellular extension (C, inset, arrow). In some cases, Salmonella-harboring CD11c+ were seen moving into the lumen (D, inset, arrow) (scale bar, 20 μm). The images are typical of 5 mice/group. Gastroenterology 2009 137, 579-587.e2DOI: (10.1053/j.gastro.2009.04.010) Copyright © 2009 AGA Institute Terms and Conditions

Figure 2 TEM micrographs show that Salmonella induced the formation of both intraepithelial cell extensions (CE) protruding into the lumen (A and B) higher magnification and the migration of leukocytes infiltrating (IC), in close contact with Salmonella (SL), between epithelial cells (E) that reached the lumen (C). Panel D shows a cell moving into the lumen with a large portion of the body already protruding from the epithelium. Arrowheads indicate intact cell junctions. Gastroenterology 2009 137, 579-587.e2DOI: (10.1053/j.gastro.2009.04.010) Copyright © 2009 AGA Institute Terms and Conditions

Figure 3 Flow cytometric quantitative and qualitative analysis of intraluminal cell population of BALB/c mice following challenge with PBS, E coli DH5α (EC D5), or Salmonella SL1344-ΔSPI1 (SL ΔSPI1). The intraluminal cells associated with GFP-labelled bacteria were analyzed using a gating strategy that combined light scatter with GFP fluorescence to exclude extracellular bacteria (A, upper panel). The relative fluorescence of the GFP-labelled E coli and Salmonella is shown (A, lower panels). The gated population of murine cells was CD11c+CX3CR1+MHCII+ but did not express either CD11b or CD8α (B). Quantitative analysis is shown in C. The total number of intraluminal DCs was significantly higher at 1.5 hours after the challenge with Salmonella and increased further after 3 hours. In contrast, no changes in the number of intraluminal DCs were seen throughout the experiment when either PBS or E coli was used. Gastroenterology 2009 137, 579-587.e2DOI: (10.1053/j.gastro.2009.04.010) Copyright © 2009 AGA Institute Terms and Conditions

Figure 4 Salmonella-bound flagellin is critical for the transepithelial migration of DCs. Both soluble flagellin (FliC) and flagellin-deficient Salmonella (SL ΔfliC ΔfljB) failed to induce transepithelial migration of DCs compared with the flagellated strain (SL ΔSPI1) (A). In addition, DC migration was absent following challenge with flagellated E coli K12 (EC K12) or Salmonella SPI1-SPI2 double mutant (ΔSPI1 ΔssrA). Flagellin-deficient Salmonella mutant was translocated across the epithelium (B, insets). Gastroenterology 2009 137, 579-587.e2DOI: (10.1053/j.gastro.2009.04.010) Copyright © 2009 AGA Institute Terms and Conditions

Figure 5 Salmonella-induced migration of DCs is MyD88-dependent. Flow cytometric analysis (see Figure 3 legend for details) of Salmonella-induced migration of CD11c+CX3CR1+ MHCII+CD11b−CD8α− DCs in the lumen of C57BL/6 mice (A). (B) Shows that migration observed in wild-type C57BL/6 mice was abolished in mice that lack the adaptor molecule MyD88. Gastroenterology 2009 137, 579-587.e2DOI: (10.1053/j.gastro.2009.04.010) Copyright © 2009 AGA Institute Terms and Conditions

Figure 6 Salmonella-induced transepithelial migration of DCs is restricted to the small intestine of BALB/c mice. Numbers of DCs/centimeter of intestine appeared to be slightly higher in the jejunum than in the ileum, but the difference was not statistically significant; in contrast, negligible number of DCs was detected in the colon (A). Furthermore, DC migration was observed in the small intestine of BALB/c mice 5 hours after oral delivery (gavage) of Salmonella (B). Gastroenterology 2009 137, 579-587.e2DOI: (10.1053/j.gastro.2009.04.010) Copyright © 2009 AGA Institute Terms and Conditions

Figure 7 The integrity of the epithelial barrier was maintained. The epithelial barrier was assessed by TEM (A, higher magnification in B), which shows intact junctions (arrowheads). Immunostaining (C) with anti-occludin antibody (blue) on a cryosection shows a typical distribution of occludin, even in presence of GFP-labelled Salmonella internalized by CD11c+ DC (red). Light microscopy also shows normal morphology of the gut mucosa following challenge with Salmonella (D). In addition, gut permeability did not increase after challenge with Salmonella as assessed by determining plasma levels of FITC-dextran (Supplementary Figure 1). Gastroenterology 2009 137, 579-587.e2DOI: (10.1053/j.gastro.2009.04.010) Copyright © 2009 AGA Institute Terms and Conditions

Figure 8 Intraluminal DCs capture Salmonella ΔSPI1. (A) Displays a luminal cell with the typical morphologic feature of a DC harboring a Salmonella bacterium (higher magnification in B). Immunostaining shows an intraluminal CD11c+DC (red) harboring 2 GFP-labelled Salmonella bacterium (C). The ability of intraluminal DCs to capture Salmonella was also proved by plating lysates of isolated gentamicin-treated luminal DCs on agar plates (D). Gastroenterology 2009 137, 579-587.e2DOI: (10.1053/j.gastro.2009.04.010) Copyright © 2009 AGA Institute Terms and Conditions

Supplementary Figure S1 Permeability of the small intestine following bacterial challenge. Plasma levels of FITC-dextran were determined to assess the permability of the gut 3 hours after challenge with Salmonella ΔSPI1 (SL ΔSPI1) that induced DC migration and E coli DH5α (EC D5) that failed to do so. No significant differences were seen, thus showing that DC migration is not linked to increased permeability of the gut. Gastroenterology 2009 137, 579-587.e2DOI: (10.1053/j.gastro.2009.04.010) Copyright © 2009 AGA Institute Terms and Conditions