Volume 122, Issue 3, Pages 667-680 (March 2002) Ubiquitous production of macrophage migration inhibitory factor by human gastric and intestinal epithelium  Christian Maaser, Lars Eckmann, Günther Paesold, Hyun S. Kim, Martin F. Kagnoff  Gastroenterology  Volume 122, Issue 3, Pages 667-680 (March 2002) DOI: 10.1053/gast.2002.31891 Copyright © 2002 American Gastroenterological Association Terms and Conditions

Fig. 1 MIF expression in human gastrointestinal mucosa. Frozen sections of normal human (A) gastric, (D) duodenal, and (G) colon mucosa as well as (J) mucosa from a colon xenograft were immunostained for MIF. Arrows in G point to lamina propria cells that also stained positive for MIF. Adjacent sections (B, E, H, and K) were stained with an isotype-matched control antibody. Little immunostaining was noted after preabsorption of MIF antibody with recombinant MIF, further confirming the specificity of immunostaining (data not shown). Corresponding sections were stained with H&E (C, F, I, and L). Sections are oriented with the lumen to the right. (Original magnification 20×.) Gastroenterology 2002 122, 667-680DOI: (10.1053/gast.2002.31891) Copyright © 2002 American Gastroenterological Association Terms and Conditions

Fig. 2 Immunoblot analysis of MIF expression in human intestine. (A) Immunoblot analysis of cell lysates (protein amount, 9 μg/lane) from noninflamed gastric, duodenal, and colon mucosa and from a human intestinal xenograft. rhMIF (20 ng) was used as a positive control. Similar results were obtained using lysates from gastric, duodenal, and colon biopsy specimens from 3 additional individuals and intestinal xenografts from 3 additional donors. (B) Primary epithelial cells were isolated from small intestinal xenograft tissue. Total protein lysates recovered from the fractions of a 10%–90% Percoll gradient were size separated on a 10%–20% acrylamide gradient gel. A representative immunoblot is shown. Fractions are labeled 1 to 12, with 1 representing the fraction with the lowest and 12 with the highest concentration of Percoll. Similar results were obtained in 4 repeated experiments. (C) To verify the characteristics of the cells in the fractions from the density gradients, several approaches were used and the data are summarized here. The localization of epithelial cells in the fractions was determined by direct cell counts of the fractions and immunoblot analysis of cell lysates of each fraction for the epithelial cell–specific marker cytokeratin 8/18. T cells were localized on the gradients by direct cell counts and immunoblot analysis of lysates of the cells in each fraction for vimentin, which is expressed by all lymphocytes but not epithelial cells. Because gradients with primary epithelial cell preparations from intestinal xenografts did not contain sufficient numbers of T cells to allow detection by immunoblot analysis, some primary cell preparations were spiked with HUT 78 human T lymphoma cells and analyzed on parallel gradients. ++ and + indicate the presence of high and medium-low numbers of the cell type, respectively; − indicates the absence in the fraction of the cell type indicated on the left. Gastroenterology 2002 122, 667-680DOI: (10.1053/gast.2002.31891) Copyright © 2002 American Gastroenterological Association Terms and Conditions

Fig. 3 Constitutive expression of MIF mRNA by colon and gastric epithelial cell lines. MIF and β-actin mRNAs were amplified by reverse-transcription PCR. In the control lane labeled “no RNA,” RNA was omitted from the reverse transcription reaction and PCR amplification. Gastroenterology 2002 122, 667-680DOI: (10.1053/gast.2002.31891) Copyright © 2002 American Gastroenterological Association Terms and Conditions

Fig. 4 Immunoblot analysis of MIF expression by human colon cell lines. Cell lysates and supernatants from confluent monolayers of Caco-2, HT-29, and T84 cells were subjected to immunoblot analysis. MIF was detected using polyclonal goat anti-human MIF antibody and enhanced chemiluminescence. rhMIF (20 ng) was loaded as a positive control. Lanes 3–5 were loaded with 9 μg/lane of total protein extract from Caco-2, HT-29, and T84 cells, respectively. Lanes 7–9 were loaded with equal volumes of concentrated supernatants from confluent monolayers of these cells. Gastroenterology 2002 122, 667-680DOI: (10.1053/gast.2002.31891) Copyright © 2002 American Gastroenterological Association Terms and Conditions

Fig. 5 Inhibition of U937 migration by epithelial cell–produced MIF. Caco-2 cell supernatant (SN) was pretreated with mouse anti-human MIF mAb (44 μg/mL), mouse IgG1 as an isotype control, or PBS. Calcein–acetoxymethyl ester (AM)–labeled U937 cells were resuspended in the indicated supernatants, SDF-1α (50 ng/mL) was added to the bottom wells, and U937 cells were added to the top wells of the chemotaxis plate. Migration of U937 cells into the filter was determined after 90 minutes. Data are shown as number of cells that migrated into the filter minus the number of randomly migrated cells, which averaged 3.25 × 105 cells/filter. Values are expressed as means ± SEM of 3 independent experiments. *Values significantly different from “no SN” controls (P < 0.01). Gastroenterology 2002 122, 667-680DOI: (10.1053/gast.2002.31891) Copyright © 2002 American Gastroenterological Association Terms and Conditions

Fig. 6 p-hydroxyphenylpyruvate isomerization activity of epithelial cell–produced MIF. (A) Caco-2 cell lysates (total MIF content, 82 ng/mL) incubated with mouse IgG1 as isotype control (■) or mouse anti-human MIF mAb (22 μg/mL) (♢) were assayed for p-hydroxyphenylpyruvate isomerization activity as described in Materials and Methods. Values are given as increase in absorbance at 330 nm. (B) Caco-2 cell lysates (■) and supernatants (♢) adjusted to contain equal concentrations of MIF (4.7 ng/mL) were assayed for p-hydroxyphenylpyruvate isomerization activity. The values are expressed as means of triplicate cultures from a representative experiment. Similar results were obtained in 3 independent experiments. Gastroenterology 2002 122, 667-680DOI: (10.1053/gast.2002.31891) Copyright © 2002 American Gastroenterological Association Terms and Conditions

Fig. 7 MIF expression in polarized Caco-2 cells. Caco-2 cells were grown on collagen-coated filter inserts. Monolayers were stained for MIF (red fluorescence; A and C) or with an isotype control antibody (B and D), counterstained with Alexa 488-conjugated phalloidin (green fluorescence), and analyzed by confocal microscopy. A and B are representative optical sections obtained at approximately one-third depth from the apical surface. C and D show confocal X-Z reconstructions. Gastroenterology 2002 122, 667-680DOI: (10.1053/gast.2002.31891) Copyright © 2002 American Gastroenterological Association Terms and Conditions

Fig. 8 Secretion of MIF, CXCL8/IL-8, and CXCL10/IP-10 by polarized Caco-2 cells. Polarized Caco-2 cells were stimulated with IL-1α (20 ng/mL) and IFN-γ (40 ng/mL) or a combination of both cytokines added to the basal and apical chambers or left unstimulated. Apical and basal chamber supernatants were harvested after 16 hours and assayed by ELISA for MIF (top), CXCL8/IL-8 (middle), and CXCL10/IP-10 (bottom). ■, Apical secretion; ● basolateral secretion. Values are mean ± SEM of 3 independent experiments. *Values significantly different from unstimulated controls in the respective compartment (P < 0.05). Gastroenterology 2002 122, 667-680DOI: (10.1053/gast.2002.31891) Copyright © 2002 American Gastroenterological Association Terms and Conditions

Fig. 9 MIF in colon lavage fluid. Colon lavage fluid from healthy patients undergoing colonoscopy for colon cancer screening was concentrated and subjected to immunoblot analysis. MIF was detected using polyclonal goat anti-human MIF antibody and enhanced chemiluminescence. rhMIF (20 ng) was loaded as a positive control. The results from 2 representative lavage fluids are shown. Similar results were obtained with samples from 4 additional patients. Gastroenterology 2002 122, 667-680DOI: (10.1053/gast.2002.31891) Copyright © 2002 American Gastroenterological Association Terms and Conditions

Fig. 10 MIF secretion and mRNA expression by Caco-2 cells infected with S. dublin. Confluent monolayers of Caco-2 cells in 6-well plates were infected with S. dublin (♦) or left uninfected as controls (▴) and subsequently incubated with media containing gentamicin. (A) Supernatants were collected at the times indicated, and MIF levels were assayed by ELISA. Values are means ± SEM of 3 independent experiments. *Values significantly different from uninfected controls at the same time point (P < 0.01). (B) Qualitative reverse-transcription PCR analysis. MIF and CXCL8/IL-8 mRNA expression as a positive control23 was determined as described in Materials and Methods. A representative analysis of 1 of 3 experiments performed is shown. (C) Lactate dehydrogenase activity in the same supernatants as in A was measured by enzymatic assay. Values are means ± SEM of 3 independent experiments. *Values significantly different from uninfected controls at the same time point (P < 0.01). Gastroenterology 2002 122, 667-680DOI: (10.1053/gast.2002.31891) Copyright © 2002 American Gastroenterological Association Terms and Conditions