Nutrient-Gene Expression Elevated Iron Status Increases Bacterial ...

Nutrient-Gene Expression

Elevated Iron Status Increases Bacterial Invasion and Survival and Alters Cytokine/Chemokine mRNA Expression in Caco-2 Human Intestinal Cells1 Susan L. Foster,*† Stephen H. Richardson† and Mark L. Failla*2 *Department of Nutrition and Foodservice Systems, The University of North Carolina at Greensboro, Greensboro, North Carolina 27402 and †Department of Microbiology and Immunology, Wake Forest University School of Medicine, Winston-Salem, North Carolina 27157 ABSTRACT Iron status affects both microbial growth and immune function. Mammalian iron homeostasis is maintained primarily by regulating the absorption of the micronutrient in the proximal small intestine. The iron concentration of the enterocyte can fluctuate widely in response to both dietary and whole body iron status, as well as in response to infections. The possibility that an enterocyte with an elevated iron concentration is more susceptible to invasion by enteric pathogens is not known. Therefore, we examined the impact of enterocyte iron status on the invasion and survival of an enteric pathogen, as well as on the levels of several cytokine and chemokine mRNAs by the host cell. The enterocyte-like Caco-2 human intestinal cell line and Salmonella enteritidis served as the models to examine the effect of iron on the host-parasite interaction. Iron status of Caco-2 cells was altered by incubation in serum-free medium supplemented with varying levels of iron. Elevated iron status of Caco-2 cells increased the efficiency of the invasion and the number of bacteria surviving in the intracellular environment. Caco-2 cells constitutively expressed transforming growth factor-␤1, interleukin-8, monocyte chemotactic protein-1, tumor necrosis factor-␣ and interleukin-1␤, and infection with S. enteritidis increased the relative quantities of all cytokine/chemokine mRNAs except interleukin-1␤. Elevated iron status of Caco-2 cells decreased the levels of cytokine/chemokine mRNAs by 25– 45% in uninfected cells. In contrast, bacterial infection was associated with a 21–95% increase in cytokine/chemokine mRNAs levels in Caco-2 cells with higher iron concentration compared with infected cells with lower iron concentration. These data support the hypothesis that elevated enterocyte iron status increases susceptibility to infection and exacerbates the mucosal inflammatory response initiated by microbial invasion by increasing cytokine/chemokine expression. J. Nutr. 131: 1452–1458, 2001. KEY WORDS: ● chemokines

● ●

iron ● Caco-2 human intestinal cells infection

●

The micronutrient iron plays a central role in many biochemical and physiological processes. Mammalian iron homeostasis is maintained primarily by regulating absorption of the metal in the proximal small intestine (1–3). During conditions of iron deficiency, the enterocyte efficiently transfers dietary iron to the plasma, whereas the enterocyte retains iron acquired from the diet when whole body iron concentration is elevated. Thus, the concentration of iron in the enterocyte can fluctuate widely in response to both dietary and whole body iron status. The iron concentration of the enterocyte is also elevated in response to infections. Decreased intestinal absorption of iron in response to infection has been observed in humans, rats and chicks (4 – 6). High intracellular concentrations of iron generally are associated with the potential to

transforming growth factor-␤1

●

cytokines

catalyze oxidative damage through the generation of free radicals (7) There also is considerable evidence that change in the levels of iron in body fluids and tissues of the host can influence the incidence and severity of microbial infections. For example, conditions that result in hyperferremia (e.g., hemochromatosis, thalassemia and leukemia) are associated with an increased risk of developing infections (8). Legionella pneumophila, a pathogen that invades alveolar macrophages, is dependent on iron availability for growth and survival. Exposure of infected macrophages to the iron chelator desferrioxamine prevents the multiplication of intracellular L. pneumophila, whereas loading the macrophage with iron stimulates the replication of the pathogen (9). Limited iron availability also inhibits the intracellular growth of other parasites, including Trypanosoma cruzi, Plasmodium falciparum and Toxoplasma gondii (10 –12). The possibility that an enterocyte with an elevated iron concentration also will be more susceptible to invasion by enteric microorganisms has not been considered previously. In addition to their central role in nutrient absorption, enterocytes play a primary role in the initiation and regulation

1 Presented in part at Experimental Biology 99, Washington, D.C. [Foster, S. L., Richardson, S. H. & Failla, M. L. (1999) Impact of Caco-2 iron status on invasion and survival of Salmonella enteritidis and cytokine/chemokine expression. FASEB J. 13: A450.15]. 2 To whom correspondence should be addressed at Department of Nutrition Human Nutrition and Food Management, The Ohio State University, 325 Campbell Hall, Columbus, OH 43210. E-mail: [email protected]

0022-3166/01 $3.00 © 2001 American Society for Nutritional Sciences. Manuscript received 24 July 2000. Initial review completed 1 November 2000. Revision accepted 29 January 2001. 1452

IRON STATUS ALTERS CYTOKINE EXPRESSION IN CACO-2 CELLS

of the intestinal mucosal immune response (13,14). Enterocytes synthesize and secrete a variety of immunoregulatory molecules (e.g., cytokines, chemokines and growth factors) and are sentinel cells in immunosurveillance, because they process antigens for presentation on their basolateral surface (15) They also synthesize other factors that participate in host defense, including bacteriolytic enzymes (e.g., lysozyme, phospholipase A2 and ␣1-antitrypsin) and antimicrobial peptides (defensins) (16). Iron status has been reported to modulate cytokine secretion. Interleukin (IL)-13 secretion is impaired in iron-deficient rats (17), whereas tumor necrosis factor (TNF)-␣ production is increased in lipopolysaccharide-stimulated monocytes from infants with iron deficiency anemia (18). Also, iron loading decreased the T helper-1/T helper-2 cytokine ratio and exacerbated the disease process in a murine model of candidiasis (19). Likewise, excessive iron status is associated with increases in IL-1␤ and TNF-␣ secretion by alveolar macrophages and transforming growth factor (TGF)-␤ gene expression in the hepatic acinar zone 1 cells of rats (20,21). The possibility that variations in cellular iron status influence cytokine and chemokine synthesis in enterocytes in response to invasion has not been examined. The objective of the present study was to examine the impact of enterocyte iron status on the invasion and survival of an enteric pathogen and the expression of cytokine and chemokine genes in response to microbial invasion. Salmonella enteritidis was used as a representative enteric pathogen, and differentiated cultures of human Caco-2 cells served as the enterocyte-like cells. MATERIALS AND METHODS Reagents. Dulbecco’s modified Eagle’s medium (DMEM), Hanks’ balanced salt solution (HBSS), Dulbecco’s phosphate-buffered saline with and without Ca2⫹ and Mg2⫹, 1 mol HEPES/L, pH 7.4, 21.6 mmol gentamicin/L, antibiotic-antimycotic (16.9 mmol penicillin, 6.86 mmol streptomycin and 27.1 ␮mol amphotericin per L), 200 mmol L-glutamine/L, trypsin-EDTA (0.25%, 1 mmol EDTA 䡠 4Na/L), fetal bovine serum (FBS) and 10 mmol minimal essential medium nonessential amino acids (NEAA)/L were purchased from GIBCO BRL (Grand Island, NY). Ferric chloride (FeCl3 䡠 6H2O) and ascorbic acid (C6H8O6) were obtained from Sigma Chemical Company (St. Louis, MO). Caco-2 cell culture. The human colon adenocarcinoma cell line Caco-2 (HTB-37) was purchased from American Type Culture Collection (Rockville, MD). Caco-2 cells were used between passages 23 and 43. Stock cultures were maintained in 75-cm2 T-flasks (Corning Science Products, Corning, NY) in complete medium in a humidified atmosphere of 95% air and 5% CO2 at 37°C. The complete medium contained basal DMEM supplemented with 25 mmol glucose, 15 mmol HEPES, 2 mmol glutamine, 100 ␮mol NEAA and 10 mL heat-inactivated FBS per L and antibiotics (169 ␮mol penicillin, 68.6 ␮mol streptomycin and 271 nmol amphotericin per L). The iron concentration of the complete culture medium was 5.3 ␮mol/L as determined by atomic absorption spectrophotometry. Stock cultures were seeded at 6700 cells/cm2 and split by treatment with trypsinEDTA (2.5 g trypsin and 1 mmol tetrasodium EDTA in Ca2⫹- and Mg2⫹-free HBSS per L) at ⬃75% confluency. For experiments examining bacterial invasion and survival, Caco-2 cells were cultured in 24-well culture dishes (Falcon, Lincoln Park, NJ). Monolayers for invasion, survival and quantification of mRNA were prepared by seeding 5.0 ⫻ 104 cells into 1.0 mL of

3 Abbreviations used: CFU, colony-forming units; DMEM, Dulbecco’s modified Eagle’s medium; ELISA, enzyme-linked immunosorbent assay; FBS, fetal bovine serum; GM-CSF, granulocyte macrophage-colony stimulating factor; IFN, interferon; IL, interleukin; MCP-1, monocyte chemotactic protein-1; NEAA, nonessential amino acids; RPA, ribonuclease protection assay; TGF-␤1, transforming growth factor-␤1; TNF-␣, tumor necrosis factor-␣.

1453

complete medium per well of a 24-well culture plate. Monolayers for measuring Caco-2 iron concentration (ferritin) were prepared by seeding 2.5 ⫻ 105 cells in 2.5 mL of complete medium onto each well of a 6-well culture plate. Medium was changed every 2 d and on the day before an experiment. Experiments were generally performed with cultures at 10 –12 d after monolayers reached confluency (22). The impact of degree of differentiation of Caco-2 cells on the invasion and survival of S. enteritidis was evaluated by using monolayers at d 3 versus d 12 after confluency to compare undifferentiated and differentiated cells, respectively. Differentiation was assessed using alkaline phosphatase as a marker enzyme. Alkaline phosphatase activity was quantified by monitoring the conversion of p-nitrophenyl phosphate to p-nitrophenyl at 400 nm (23). Bacterial strain and growth conditions. S. enteritidis CDC5str, a clinical isolate from a patient with gastroenteritis, was received from Virginia Miller (Department of Microbiology, University of California Los Angeles). CDC stock was stored at ⫺70°C in heart infusion broth (Difco Laboratories, Detroit, MI) containing 50% (v/v) glycerol. Preliminary experiments established the growth conditions for S. enteritidis for optimal invasion of Caco-2 cells. Cultures of bacteria grown in a roller tube apparatus (70 rpm) at 37°C for 16 –18 h were found to be the most invasive. Heart infusion broth (5 mL) was inoculated with a 100-␮L aliquot of frozen bacterial stock suspension and placed in the roller tube apparatus at 37°C overnight to provide bacteria for the standard invasion assay. An aliquot of the overnight culture was diluted 1:20 in 5 mL of fresh heart infusion broth and incubated for an additional 2 h at 37°C on the rotator. Growth curves verified that the bacteria were in mid-log growth phase. The number of bacteria in the inoculum was determined by turbidity readings at 600 nm and confirmed by plating dilutions onto Columbia colistin– nadidixic acid agar and quantifying the number of colony-forming units (CFU) per unit volume of inoculum (5.0 ⫻ 107 ⫾ 0.8 ⫻ 107 CFU/mL). Invasion assay. The invasion assay was a modification of the procedure developed by Elsinghorst et al. (24) as modified by Tatera and Metcalf (25). Bacteria were grown to an A600 of 0.5 (midlogarithmic phase), collected through centrifugation and washed in PBS before resuspension to 8.3 ⫻ 1010 bacteria/L. The composition of the test medium was DMEM containing 25 mmol glucose, 2 mmol L-glutamine, 0.1 mmol NEAA, 15 mmol HEPES and 20 mL FBS per L. Caco-2 monolayers were washed three times with PBS before the addition of 300 ␮L of bacterial suspension containing 5.0 ⫻ 107 bacteria to the monolayers (100 bacteria per Caco-2 cell). Differentiated Caco-2 monolayers contained 5.0 ⫻ 105 ⫾ 2.5 ⫻ 104 cells as determined by trypsinization of monolayers and counting of the cell number. Cultures were incubated at 37°C for 2 h. Medium was removed, and monolayers were washed three times with HBSS. Medium (1 mL) containing gentamicin (216 ␮mol/L) was added to each monolayer and incubated for an additional 2 h at 37°C to kill residual extracellular bacteria. The viability of Caco-2 cells in cultures of control and infected cells was ⱖ95% as assessed by trypan blue exclusion. After treatment with gentamicin, each monolayer was washed four times with HBSS, and intracellular bacteria were released by adding 200 ␮L Triton X-100 (10 g/L). Deionized water (800 ␮L) was added to each monolayer after 10 min, and the sample was pipetted repeatedly for complete lysis of the cells. Aliquots of cell lysate were diluted in saline and plated onto Columbia colistin– nadidixic acid agar supplemented with 170 mmol NaCl/L to enumerate the number of internalized bacteria. Survival assay. Monolayers were infected with bacteria as described earlier. After the 2-h gentamicin treatment, the monolayer was washed six times with PBS. Fresh test medium (1 mL) containing 108 ␮mol gentamicin/L was added, and the monolayers were incubated at 37°C for 22 h. Monolayers were washed three times with PBS. Surviving intracellular bacteria present 24 h after the initiation of invasion were quantified as described earlier. There was no evidence that infection killed the Caco-2 cells, because the monolayer remained confluent and viability exceeded 95% as assessed by trypan exclusion. Alteration of iron concentration of Caco-2 cells. The iron concentration of the Caco-2 cells was altered by incubating monolayers in serum-free medium containing varying iron levels (26). Iron was

FOSTER ET AL.

1454

presented to Caco-2 cells as ferrous ascorbate to increase its solubility, thereby facilitating cellular accumulation (27). A stock solution containing 10 mmol FeCl3 䡠 6H2O and 200 mmol ascorbic acid per L in 1 mmol HCl/L was prepared fresh for each experiment. Aliquots were added to serum-free DMEM to prepare solutions with either 0, 5 or 50 ␮mol Fe/L and either 0, 100 or 1000 ␮mol ascorbate/L, respectively, for addition to monolayers of Caco-2 cells. Because the basal concentration of iron in basal DMEM was 1.8 ␮mol/L as determined by atomic absorption spectrophotometry, the final iron concentration of the serum-free test medium was either 1.8, 6.8 or 51.8 ␮mol/L. The effect of ascorbate (100 or 1000 ␮mol/L) alone on the uptake of endogenous Fe by Caco-2 cells was also examined by adding appropriate aliquots of a stock solution of ascorbic acid (200 mmol/L) in dilute HCl (1 mmol/L). Intracellular ferritin. Intracellular iron concentration was estimated by quantifying ferritin protein using a two-stage enzyme-linked immunosorbent assay (ELISA) (Ferritin Kit; Spectro, Ramco, TX). Cells from a 6-well culture plate were pooled for each ferritin determination. Cells were collected in 1 mL of cold 150 mmol NaCl/L, pelleted by centrifugation (1000 ⫻ g for 5 min) and washed twice with cold PBS before resuspension in 500 ␮L PBS. The cell suspension was sonicated for 30 s at 10-s pulse intervals with a Branson Sonifier 250 (VWR Scientific, Atlanta, GA). Aliquots of the Caco-2 cell lysates were heated at 70°C for 5 min to destroy heat-labile proteins that might interfere with the ELISA through nonspecific binding. The linear range of the ferritin assay was 0 –200 ␮g/L with a sensitivity of 6.0 ␮g/L. The optical density was determined at 490 nm with a plate reader (Molecular Devices, Sunnyvale, CA). The ferritin concentration of the Caco-2 cell lysates was determined from a calibration curve prepared with known quantities of human ferritin and is expressed as nanograms of ferritin per milligram of cell protein. The protein concentration of the Caco-2 cell lysates was determined by a modified Lowry assay using bovine serum albumin as the standard (28). Ribonuclease protection assay. A multiprobe ribonuclease protection assay (RPA) (Riboquant; PharMingen, San Diego, CA) was used to detect and quantify cytokine/chemokine mRNAs. RNA was extracted from 12-d postconfluent Caco-2 cells using the acid guanidinium thiocyanate-phenol-chloroform method of Chomczynski and Sacchi (29). Caco-2 RNA (20 ␮g) was hybridized to 32P-labeled antisense RNA probes. The probes were synthesized by in vitro transcription from a DNA template using T7 polymerase. RNA-RNA hybrids were treated with RNase to degrade single-stranded RNA and excess probe. The radiolabeled probe and the RNase-protected fragments were separated by denaturing gel electrophoresis and quantified using a PhosphorImager (445SI; Molecular Dynamics) with ImageQuant software. After subtraction of the background, the intensity of the appropriately sized protected fragments was used as the indicator for the levels of mRNAs in the original sample. The amount of mRNA per test sample was normalized by considering the quantity of the GAPDH probe in each lane. Because TGF-␤1 mRNA was present at a higher level than that of the other cytokine/chemokine mRNAs in uninfected cells maintained under standard conditions (i.e., complete DMEM with 100 mL FBS/L), TGF-␤1 mRNA in these cells was assigned an arbitrary value of 100 relative units. The relative quantities of the cytokine/chemokine mRNAs for all test conditions are expressed as a percentage of the basal level of TGF-␤1 mRNA. Statistical analysis. Data were analyzed by ANOVA using SigmaStat software (version 1.0; Jandel Scientific, Sausilito, CA). The Student-Newman-Keuls test for multiple comparisons was used to determine significant differences (P ⱕ 0.05) between treatment groups for evaluating the effects of iron status and infection versus no infection. Results are presented as mean ⫾ SEM. Invasion and survival experiments were repeated at least three times. For each experimental condition, three wells were used, and triplicate colony counts were performed for each well. Data for ferritin concentration represent the average of three to six ELISA analyses. RPA data represent results from RNA isolated from five separate infection experiments.

RESULTS Cell concentration of ferritin (Fe) varied by altering level of Fe in medium. Differentiated cultures (12 d) of Caco-2

cells were exposed to serum-free medium supplemented with low (0 ␮mol/L), moderate (5 ␮mol/L) or high (50 ␮mol/L) concentrations of iron and a 20-fold excess of ascorbate for 24 h to vary cell Fe concentration. Iron supplementation did not adversely affect general cellular morphology, viability or protein concentration of the monolayers (data not shown). Intracellular ferritin concentration was used as a marker of iron status. The ferritin concentration of 12-d postconfluent cultures maintained under standard conditions (i.e., DMEM with 100 mL FBS and 5.3 ␮mol Fe per L) was 5.7 ⫾ 0.5 ng/mg protein. Serum-free cultures incubated without supplemental iron (0 ␮mol/L Fe) contained approximately twice as much ferritin protein (12.2 ⫾ 2.2 ng/mg protein) as those maintained in the presence of 100 mL FBS/L. Cell ferritin concentration increased in proportion to medium iron concentration. Cell ferritin concentration after the addition of medium supplemented with 5 ␮mol Fe and 100 ␮mol ascorbate per L was 15-fold (186.3 ⫾ 42.2 ng/mg) higher than that in unsupplemented, serum-free cultures of Caco-2 cells. Exposure of cultures to 50 ␮mol Fe and 1000 ␮mol ascorbate per L further increased cell ferritin to almost 50-fold (594.5 ⫾ 144.5 ng/mg) that of the unsupplemented group (Fig. 1A). These data suggest that the availability of iron for the cell is greater when the metal is presented as a low-molecular-weight chelate that can be reduced by ascorbate than as transferrin-Fe(III) complex. Because ascorbate is a known promoter of cellular iron acquisition, the influence of ascorbate itself on ferritin concentration was examined. Cell ferritin increased in differentiated (12-d postconfluent) Caco-2 cultures incubated in medium with ascorbate but without supplemental iron. The ferritin concentration of Caco-2 monolayers exposed to 100 and 1000 ␮mol ascorbate/L increased 289% (47.5 ⫾ 12.1 ng/mg protein) and 373% (57.7 ⫾ 15.0 ng/mg protein), re-

FIGURE 1 Elevated iron status of differentiated (12-d postconfluent) cultures of Caco-2 cells increases the efficiency of invasion and the number of intracellular Salmonella enteritidis at 24 h after infection. A. Effect of medium Fe concentration on ferritin levels in differentiated Caco-2 cultures. B. Impact of Caco-2 Fe concentration on invasion and survival of S. enteritidis. Ferritin concentration, invasion, and survival were determined as described in Materials and Methods. Values are mean ⫾ SEM, n ⫽ 6 independent experiments for ferritin measurements, n ⫽ 9 for invasion studies and n ⫽ 3 for survival studies. Within each panel, columns with different letters are significantly different, P ⱕ 0.05.


spectively, compared with the unsupplemented control. The increased cellular ferritin concentration in response to medium ascorbate concentration demonstrated that the reducing agent facilitated cellular acquisition of basal (i.e., iron present in the culture medium), as well as supplemental, iron from the culture medium. Impact of Fe status of Caco-2 cells on invasion and survival of CDC S. enteritidis. Initial studies evaluated the ability of S. enteritidis to invade and subsequently survive in monolayers of differentiated Caco-2 cells. The number of bacteria (CFU/mL) present in Caco-2 cells 2 h after infection was used as a measure of invasion. The number of bacteria internalized in Caco-2 monolayers 24 h after infection was used as an index of survival and multiplication. S. enteritidis exhibited the ability to infect and survive in Caco-2 cells cultured under normal culture conditions (i.e., DMEM with 100 mL FBS/L). Less than 1% (2.75 ⫻ 105 ⫾ 4.5 ⫻ 104 CFU/mL) of the initial bacterial inoculum was internalized by 2 h after introduction of the bacterium into Caco-2 cultures. The bacterial load present in cells 22 h after the removal of extracellular bacteria was 7.75 ⫻ 104 ⫾ 7.5 ⫻ 103 CFU/mL of the inoculum, suggesting that some killing had occurred. The impact of cell iron status on S. enteritidis invasion and survival in Caco-2 cells was examined next (Fig. 1B). The efficiency of bacterial invasion was 57 and 68% greater (P ⱕ 0.05) in cultures previously incubated in medium with moderate (5 ␮mol Fe/L) and high (50 ␮mol Fe/L) iron concentrations, respectively, compared with Caco-2 cultures incubated without supplemental iron (Fig. 1B). The number of viable intracellular bacteria at 22 h after invasion was significantly (P ⬍ 0.05) higher in all cultures maintained in serumfree medium than that in medium with 100 mL FBS/L. Also, the number of intracellular bacteria was 81 and 131% greater in cultures exposed to serum-free medium containing moderate (5 ␮mol Fe/L) and high (50 ␮mol Fe/L) concentrations of iron, respectively, compared with cultures maintained in serum-free medium without supplemental iron (Fig. 1B). The increased survival of bacteria in Caco-2 cells with moderate and high iron status suggests an adverse effect of elevated iron status on the killing mechanisms of the host cell and/or increased multiplication of bacteria in an iron-enriched intracellular environment. Influence of the degree of Caco-2 differentiation on the impact of Fe status on the invasion and survival of S. enteritidis. The influence of cellular differentiation on iron uptake by Caco-2 cells has been previously shown (30). Differentiation was assessed using alkaline phosphatase as a marker enzyme. Alkaline phosphatase activity was detected 3 d [0.93 ⫾ 0.6 nmol/(min 䡠 mg)] after cells reached confluency, increased by 6 d [3.2 ⫾ 1.6 nmol/(min 䡠 mg)] and attained maximal levels at 11–18 d [10.5 ⫾ 0.5 nmol/(min 䡠 mg)]. Therefore, confluent cultures were used as undifferentiated and differentiated cells 3 and 12 d after reaching confluency, respectively. Ferritin concentration of 3-d undifferentiated cultures (6.6 ⫾ 0.78 ng/mg protein) was similar to that in 12-d cultures (5.7 ⫾ 0.45 ng/mg protein) maintained under standard conditions. Three-day cultures incubated in serum-free medium without supplemental iron (0 ␮mol Fe/L) contained twice as much ferritin protein (11.3 ⫾ 1.2 ng/mg protein) as those maintained in serum-containing medium. The ferritin concentration of undifferentiated Caco-2 cultures incubated with medium containing 5 ␮mol Fe/L supplemented with 100 ␮mol ascorbate/L increased 20-fold (239.4 ⫾ 93.0 ng/mg protein) compared with that of unsupplemented cultures (Fig. 2A). Exposure of 3-d postconfluent cultures to medium containing 50 ␮mol Fe and 1000 ␮mol ascorbate per L increased

1455

FIGURE 2 Invasion and survival of Salmonella enteritidis in undifferentiated (3-d postconfluent) cultures of Caco-2 cells is not altered by iron status. A. Effect of medium iron concentration on ferritin levels in undifferentiated (3-d postconfluent) Caco-2 cultures. B. Impact of Caco-2 iron concentration on invasion and survival of S. enteritidis. Ferritin concentration, bacterial invasion and survival were measured as described in Materials and Methods. Values are mean ⫾ SEM, n ⫽ 3 independent observation for each characteristic. Within each panel, columns with no letters in common are significantly different, P ⱕ 0.05.

cell ferritin 100-fold (1315.0 ⫾ 170.2 ng/mg protein) relative to that of the unsupplemented group (Fig. 2A). Invasion and survival of S. enteritidis were significantly (P ⱕ 0.05) lower in undifferentiated than in differentiated cultures of Caco-2 cells maintained in medium with 100 mL FBS/L. The number of inoculated bacteria that invaded differentiated cultures of Caco-2 cells (1.55 ⫻ 105 ⫾ 0.23 ⫻ 105 CFU/mL) was greater (P ⱕ 0.05) than that of undifferentiated cells (5.55 ⫻ 104 ⫾ 0.50 ⫻ 104 CFU/mL). The number of viable bacteria present in Caco-2 cells at 22 h after invasion was twice that in differentiated Caco-2 cells as in undifferentiated cells (2.14 ⫻ ⫾ 0.32 ⫻ 105 versus 1.09 ⫻ 105 ⫾ 0.27 ⫻ 105 CFU/mL, respectively; P ⱕ 0.05). The iron status of undifferentiated Caco-2 cells maintained overnight in serumfree medium did not influence bacterial invasion and survival (Fig. 2B). In summary, differentiated Caco-2 cells with elevated iron status were more susceptible to invasion by S. enteritidis than were undifferentiated cultures. Likewise, bacterial survival was increased in differentiated Caco-2 cells. Analysis of cytokine/chemokine gene expression by use of multiprobe RPA. The expression of IL-8, IL-1␤, TGF-␤1, TNF-␣, granulocyte macrophage-colony stimulating factor (GM-CSF) and monocyte chemotactic protein-1 (MCP-1) genes was examined and quantified using a multiprobe RPA. These cytokine/chemokines have been shown to be produced by Caco-2 cells (13,14,31) and have a well-documented role in modulating the inflammatory response in the intestine. IL-8, MCP-1, GM-CSF, IL-1␤ and TNF-␣ are proinflammatory mediators, whereas TGF-␤1 exhibits anti-inflammatory properties. TGF-␤1, IL-8, MCP-1, TNF-␣ and IL-1␤ mRNAs were present in differentiated cultures of Caco-2 cells maintained in medium with FBS (Fig. 3A). In contrast, GM-CSF mRNA was not expressed. The relative quantities of mRNA for the tested cytokines/chemokines in uninfected Caco-2 cells maintained under standard culture conditions were TGF-␤1 ⬎ IL-8

1456

FOSTER ET AL.

TABLE 1 Relative quantities of cytokine mRNAs in uninfected and infected cultures of differentiated Caco-2 cells maintained in serum-containing medium as determined using a multiprobe RNase protection assay1 Cytokine

Uninfected

Infected

2TGF-␤12

100.0 ⫾ 0.0 78.2 ⫾ 15.5 54.8 ⫾ 8.3 43.8 ⫾ 9.6 28.8 ⫾ 2.3 ND

490.2 ⫾ 57.73 194.2 ⫾ 30.63 87.2 ⫾ 11.73 42.2 ⫾ 5.9 53.8 ⫾ 1.033 ND

IL-8 MCP-1 IL-1␤ TNF-␣ GM-CSF

1 Values are means ⫾ SEM, n ⫽ 5. Differentiated (12-d postconfluent) cultures that had been maintained in DMEM with 100 mL fetal bovine serum/L were incubated with Salmonella enteritidis at a multiplicity of infection of 100 bacteria per Caco-2 cell for 2 h. Monolayers were washed three times with PBS before the addition of medium containing 216 ␮mol gentamicin/L for 2 h to kill extracellular bacteria. Monolayers were then washed with PBS, and cells were collected for RNA extraction and quantification of cytokine/chemokine mRNA by multiprobe ribonuclease protection assay as described in Materials and Methods. 2 TGF-␤1, transforming growth factor; IL, interleukin; MCP-1, monocyte chemotactic protein-1; TNF-␣, tumor necrosis factor-␣; GMCSF, granulocyte macrophage-colony stimulating factor; DMEM, Dulbecco’s modified Eagle’s medium. ND, not detected. 3 Relative quantities are significantly different from the uninfected controls, P ⱕ 0.05.

FIGURE 3 Representative gels of cytokine/chemokine mRNAs in differentiated (12-d postconfluent) cultures of Caco-2. RNA samples were isolated from uninfected and infected cultures differentiated Caco-2 cells that had been maintained in either medium with 100 mL fetal bovine serum/L (A) or serum-free medium containing either 0, 5 or 50 ␮mol supplemented Fe/L (B). Target RNA (20 ␮g) was hybridized to 32 P-labeled antisense RNA probes and in vitro transcribed from a DNA template. Samples were treated with RNase to degrade singlestranded RNA and excess probe. The probe and target RNAs were resolved by gel electrophoresis, and the protected probes were visualized using phosphorimaging.

icantly (P ⱕ 0.05) greater than those in infected cells with low iron status (Table 2). IL-1␤ mRNA was not affected by iron status in either uninfected or infected cells.

TABLE 2 Effects of medium iron concentration on relative quantities of cytokine mRNAs in uninfected and infected Caco-2 cells as determined using a multiprobe RNase protection assay1 Iron in culture medium,2 ␮mol/L Cytokine

⬎ MCP-1 ⬎ IL-1␤ ⬎ TNF-␣ (Table 1). Infection with S. enteritidis enhanced the expression of all cytokines except IL-1␤. TNF-␣, MCP-1 and IL-8 mRNA expression increased 87, 59 and 149%, respectively, whereas TGF-␤1 expression increased 490% in response to infection (Table 1). These results are in agreement with previous reports that bacterial invasion increases cytokine gene expression. The relative quantities of cytokine/chemokine mRNAs present in uninfected Caco-2 cells cultured in serum-free medium without supplemental iron (Table 2) were very similar to those in Caco-2 cells cultured in serum-containing medium (Table 1). The impact of cellular iron status on the relative quantities of cytokine mRNAs was examined in uninfected and infected Caco-2 cells maintained in serum-free medium. The levels of TGF-␤1 and MCP-1 mRNAs in uninfected Caco-2 cells with high iron status were significantly (P ⱕ 0.05) less than those in unsupplemented controls (Fig. 3B, Table 2). In contrast, levels of TGF-␤1, IL-8, MCP-1 and TNF-␣ mRNAs in infected Caco-2 cells with high iron concentration were signif-

Uninfected TGF-␤13 IL-8 MCP-1 IL-1␤ TNF-␣ Infected TGF-␤1 IL-8 MCP-1 IL-1␤ TNF-␣

0

5

115.4 ⫾ 18.9a 78.4 ⫾ 20.4 45.4 ⫾ 9.0a 35.1 ⫾ 9.8 22.6 ⫾ 3.9

97.4 ⫾ 13.3a,b 73.8 ⫾ 16.5 44.2 ⫾ 7.6a 33.2 ⫾ 6.2 21.6 ⫾ 1.9

511.0 ⫾ 72.9b* 164.0 ⫾ 24.5b* 88.6 ⫾ 11.2b* 41.4 ⫾ 4.2 47.2 ⫾ 6.0b*

542.0 ⫾ 72.9b* 214.0 ⫾ 27.4a* 85.0 ⫾ 8.9b* 40.4 ⫾ 3.1 71.4 ⫾ 6.3ab*

50 65.8 ⫾ 56.2 ⫾ 30.4 ⫾ 26.4 ⫾ 15.2 ⫾

0.0b 7.6 3.8b 2.8 1.8

833.0 ⫾ 162.0a* 236.0 ⫾ 40.2a* 110.0 ⫾ 13.2a* 50.4 ⫾ 4.0 92.2 ⫾ 9.6a*

1 Values are means ⫾ SEM, n ⫽ 5. The indicated values represent the normalized intensity of bands compared with that of TGF-␤1 mRNA in uninfected cultures of Caco-2 cultured in serum containing medium. Means within a row not sharing a superscript letter are significantly different, P ⱕ 0.05. * Relative quantity of cytokine/chemokine mRNA in infected cells differs significantly from that in uninfected cells exposed to medium with identical iron concentration, P ⱕ 0.05. 2 Serum-free Dulbecco’s modified Eagle’s medium supplemented with either 0, 5 or 50 ␮mol Fe/L and 0, 100 or 1000 ␮mol ascorbate/L, respectively. 3 TGF-␤1, transforming growth factor; IL, interleukin; MCP-1, monocyte chemotactic protein-1; TNF-␣, tumor necrosis factor-␣.


DISCUSSION Caco-2 cells respond to elevated levels of exogenous iron by increasing the uptake and storage of the metal in ferritin protein (22,26,32). In this study, we observed that Caco-2 accumulation of exogenous iron as ferritin was affected by the degree of cellular differentiation. The ferritin concentration of undifferentiated cells was 2-fold higher than that in differentiated cells incubated in a high iron medium. This suggests that the uptake of iron by a transferrin-independent pathway is greater in undifferentiated Caco-2 cells. Perhaps this is due to the lack of a fully developed homeostatic mechanism for the regulation of iron uptake and efflux. Han et al. (30) also reported that transferrin iron uptake from the basolateral surface of undifferentiated Caco-2 cells was 2- to 4-fold greater than that of fully differentiated cells. The potential impact of iron status of the host cell on the ability of S. enteritidis to invade and survive was examined using fully differentiated (12-d postconfluent) Caco-2 cultures. Efficiency of invasion and the number of bacteria surviving the intracellular environment increased when the iron level of the host cell was elevated. This suggests that the elevated iron status of the enterocyte increases the risk of enteric infections. One likely explanation for this increased virulence is the greater availability of iron for the growth of S. enteritidis. Byrd and Horwitz (9) reported that the survival and growth of Legionella pneumophila, an intracellular pathogen that invades alveolar macrophages, also were dependent on iron availability. Iron loading of macrophages increased intracellular bacterial growth, whereas exposure to the intracellular iron chelator desferrioxamine inhibited their growth. Desferrioxamine also inhibits the growth of other intracellular parasites, including T. cruzi and P. falciparum (10,11). Dimier and Bout (12) showed that this phenomenon was not limited to treatment with chelators. Replication of T. gondii in primary rat hepatocytes was inhibited by the ability of interferon (IFN)-␥ to decrease iron availability. The RPA was used to quantitatively examine the potential relationships among cell iron status, bacterial infection and cytokine/chemokine gene expression in differentiated Caco-2 cells. TGF-␤1 mRNA was the most abundant of the cytokines and chemokines that we examined and was followed in abundance sequentially by IL-8, MCP-1, TNF-␣ and IL-1␤ mRNAs. TGF-␤1 is a pluripotent cytokine with a variety of functions, including epithelial cell growth, wound repair, barrier function and regulation of cytokine secretion by lymphocytes (33). TGF-␤1–mediated repair of the invaded intestinal epithelium likely restores barrier function and limits further microbial invasion. Others have reported that TGF-␤1 is expressed constitutively by human and rat intestinal cell lines (31,34,35). Jung et al. (31) also found that the relative quantity of TGF-␤1 mRNA and secreted protein was greater than that of the other cytokines/chemokines (IL-8, MCP-1, GMCSF and TNF-␣) in Caco-2 cells and that expression of the chemokines MCP-1 and IL-8 was greater than that of the proinflammatory cytokines TNF-␣, GM-CSF and IL-1 (31). Our results are quite similar to observations of Jung et al. (31), except that we did not detect GM-CSF mRNA. This difference may be related to degree of differentiation, because our studies were conducted with differentiated cultures of Caco-2 and the previous investigators (31) examined expression in nondifferentiated cultures. The up-regulation of proinflammatory cytokine/chemokine expression and secretion in response to bacterial invasion has been previously reported for Caco-2 cells. Caco-2 cells infected with Salmonella dublin or treated with IL-1␤ had in-

1457

creased the expression of IL-8, MCP-1, GM-CSF and TNF-␣ (31). These changes required invasion by viable cells, because the levels of the cytokine/chemokine mRNAs were not altered when noninvasive E. coli or lipopolysaccharide was added to cultures. We also found that the levels of TGF-␤1, IL-8, MCP-1 and TNF-␣ mRNAs increased several-fold when Caco-2 cells were infected by S. enteritidis. In contrast, the level of IL-1␤ mRNA was not changed and GM-CSF mRNA remained undetectable in the infected cells. Others also have reported that the levels of IL-1␣ and IL-1␤ mRNAs in intestinal epithelial cell lines are not increased by stimuli such as IL-1, IL-6, TNF-␣ or TGF-␤1 (31,36). Similarly, IL-1 has not been detected in isolated intestinal epithelial cells (34,36,37). Increased levels of TGF-␤1 mRNA in Caco-2 cells in response to bacterial invasion have not been previously reported. Recent studies suggest that a balance of proinflammatory and anti-inflammatory cytokines in the intestinal epithelium is necessary to prevent intestinal inflammation. The increased level of TGF-␤1, relative to the expression of the other proinflammatory cytokines/chemokines may provide the necessary balance between the proinflammatory and anti-inflammatory cytokines/chemokines. Elevated iron status in uninfected Caco-2 cells was associated with a decline in the relative quantities of TGF-␤1 and MCP-1 mRNAs. In contrast, elevated iron status enhanced the expression of TGF-␤1, IL-8, MCP-1 and TNF-␣ mRNAs in differentiated Caco-2 cells infected with S. enteritidis. This exaggerated response may have occurred because the intracellular bacterial burden was greater in Caco-2 cells with elevated iron status. Altered cytokine expression and action in response to changes in cellular iron status have been previously noted. Exposure of THP-1 cells to several forms of iron attenuated IFN-induced responses such as production of neopterin, degradation of tryptophan and expression of MHC class II antigens (38). Conversely, the addition of desferrioxamine increased these IFN-␥–mediated events. Likewise, the production and activity of TNF-␣ were decreased in iron-loaded macrophages (39,40). TNF-␣ mRNA levels also were suppressed in response to the addition of hemin to cultures of the human monocytic cell line THP-1 (41). Additional studies are necessary to determine whether the altered cytokine/chemokine expression associated with elevated iron concentration increases susceptibility of the host to infection and affects the extent of inflammation in response to irritants. It also will be important to consider the possible effect of ascorbic acid that was present in the iron-supplemented cultures on the expression of cytokines and chemokines in future studies. Bacterial invasion and survival were affected by the degree of differentiation of Caco-2 cells. S. enteritidis invaded and survived more efficiently in differentiated Caco-2 cells than in undifferentiated cells. Binding of bacteria to enterocytes and subsequent entry into the cells involves the interaction of specific host cell receptors and ligands. In vivo studies have shown that S. typhimurium invades the brush border membrane of epithelial cells on the intestinal villi in guinea pigs and rabbits (42,43). Likewise, several Salmonella species (44), Vibrio cholera (45), enteropathogenic E. coli (46) and enterotoxigenic E. coli (47) only invade the brush border surface of differentiated intestinal cells. Antimicrobial activities (i.e., bacteriolytic activity and expression of antimicrobial peptides) of host cells against internalized pathogens also are associated with degree of differentiation. For example, Bernet-Cambert et al. (16) observed that undifferentiated INT404 and T84 intestinal cells lacked the ability to lyse bacteria or express the antimicrobial peptides PR-39 and cecropin P1. In summary, elevated iron status increases the efficiency of

FOSTER ET AL.

1458

invasion and survival of S. enteritidis in differentiated Caco-2 cells. Cytokine/chemokine mRNA expression was attenuated by high iron levels in uninfected Caco-2 cells, whereas expression was increased in infected Caco-2 cells. Collectively, these results support the hypothesis that elevated iron status of the mature enterocyte increases its susceptibility to infection. LITERATURE CITED 1. Bothwell, T. H. (1995) Overview and mechanisms of iron regulation. Nutr. Rev. 53: 237–245. 2. Skikne, B. &. Baynes, R. D (1994) Iron absorption. In: Iron Metabolism in Health and Disease (Brock, J. H., Halliday, J. W., Pippard, M. J. & Powell, L. W., eds.), pp. 151–187. W. B. Saunders, London, U.K. 3. Conrad, M. E., Umbreit, J. N., & Moore, E. G. (1993) Regulation of iron absorption: proteins involved in duodenal mucosal uptake and transport. J. Am. Coll. Nutr. 12: 720 –728. 4. Beresford, C. H., Neale, R. J., & Brooks, O. G. (1971) Iron absorption and pyrexia. Lancet 1: 568 –572. 5. Dubach, R., Callender, T. E. & Moore, C. V. (1948) Absorption of radioactive iron in patients with fever and with anemia of varied etiology. Blood 3: 526 –542. 6. Cartwright, G. Lauritsen, M. A., Jones, P. J., Merrill, I. M. & Wintrobe, M. M. (1946) The anemia of infection: hypoferremia, hypercupremia and alterations in porphyrin metabolism in patients. J. Clin. Invest 25: 65– 80. 7. Weinberg, E. D. (1990) Cellular iron metabolism in health and disease. Drug Metab. Rev. 22: 531–579. 8. Weinberg, E. D. (1984) Iron withholding: a defense against infection and neoplasia. Physiol. Rev. 64: 65–102. 9. Byrd, T. F. & Horwitz, M. A. (1989) Interferon gamma-activated human monocytes downregulate transferrin receptors and inhibit the intracellular multiplication of Legionella pneumophila by limiting the availability of iron. J. Clin. Invest 83: 1457–1465. 10. Loo, V. G. & Lalonde, R. G. (1984) Role of iron in intracellular growth of Trypanosoma cruzi. Infect. Immunol. 45: 726 –730. 11. Raventos-Suarez, C., Pollack, S., & Nagel, R. L. (1982) Plasmodium falciparum: inhibition of in vitro growth by desferrioxamine. Am. J. Trop. Med. Hyg. 31: 919 –922. 12. Dimier, I. H. & Bout, D. T. (1998) Interferon-gamma-activated primary enterocytes inhibit Toxoplasma gondii replication: a role for intracellular iron. Immunology 94: 488 – 495. 13. McGee, D. W. (1999) Inflammation and mucosal cytokine production. In: Mucosal Immunity (Ogra, P. L., Mestecky, J., Lamm, M., Strober, W. & Bienenstock, J, eds.), pp. 559 –573. Academic Press, New York, NY. 14. Kagnoff, M. F. & Eckmann, L. (1997) Epithelial cells as sensors for microbial infection. J. Clin. Invest 100: 6 –10. 15. Christ, A. D. & Blumberg, R. S. (1997) The intestinal epithelial cell: immunological aspects. Springer Semin. Immunopathol. 18: 449 – 461. 16. Bernet-Camard, M. F., Coconnier, M. H., Hudault, S., & Servin, A. L. (1996) Differentiation-associated antimicrobial functions in human colon adenocarcinoma cell lines. Exp. Cell Res. 226: 80 – 89. 17. Heylar, L. &. Sherman, A. R. (1987) Iron deficiency and interleukin-1 production by rat leukocytes. Am. J. Clin. Nutr. 46: 346 –352. 18. Munoz, C., Olivares, M., Schlesinger, L., Lopez, M., & Letelier, A. (1994) Increased in vitro tumour necrosis factor-alpha production in iron deficiency anemia. Eur. Cytokine Netw. 5: 401– 404. 19. Mencacci, A., Cenci, E., Boelaert, J. R., Bucci, P., Mosci, P., Fe, D. C., Bistoni, F., & Romani, L. (1997) Iron overload alters innate and T helper cell responses to Candida albicans in mice. J. Infect. Dis. 175: 1467–1476. 20. O’Brien-Ladner, A. R., Blumer, B. M., & Wesselius, L. J. (1998) Differential regulation of human alveolar macrophage-derived interleukin-1beta and tumor necrosis factor-alpha by iron. J. Lab. Clin. Med. 132: 497–506. 21. Houglum, K., Bedossa, P., & Chojkier, M. (1994) TGF-beta and collagen-alpha 1 (I) gene expression are increased in hepatic acinar zone 1 of rats with iron overload. Am. J. Physiol 267: G908 –G913. 22. Han, O., Failla, M. L., Hill, A. D., Morris, E. R., & Smith, J. C., Jr. (1994) Inositol phosphates inhibit uptake and transport of iron and zinc by a human intestinal cell line. J. Nutr. 124: 580 –587. 23. Garen, A., & Levinthal, C. (1960) A fine structure genetic and chemical study of the enzyme alkaline phosphatase of E. coli. I. Purification and characterization of alkaline phosphatase. Biochim. Biophys. Acta 38: 470 – 483. 24. Elsinghorst, E. A., Baron, L. S., & Kopecko, D. J. (1989) Penetration of

human intestinal epithelial cells by Salmonella: molecular cloning and expression of Salmonella typhi invasion determinants in Escherichia coli. Proc. Natl. Acad. Sci.USA 86: 5173–5177. 25. Tartera, C. & Metcalf, E. S. (1993) Osmolarity and growth phase overlap in regulation of Salmonella typhi adherence to and invasion of human intestinal cells. Infect. Immunol. 61: 3084 –3089. 26. Alvarez-Hernandez, X., Nichols, G. M., & Glass, J. (1991) Caco-2 cell line: a system for studying intestinal iron transport across epithelial cell monolayers. Biochim. Biophys. Acta 1070: 205–208. 27. 27. Han, O., Failla, M. L., Hill, A. D., Morris, E. R., & Smith, J. C., Jr. (1995) Ascorbate offsets the inhibitory effect of inositol phosphates on iron uptake and transport by Caco-2 cells. R. Soc. Exp. Biol. Med. 210: 50 –56. 28. Nerurkar, L. K., Marino, P. A. & Adams, D. O. (1981) Quantitation of selected intracellular and secreted hydrolases of macrophages. In: Manual of Macrophage Methodology (Hersowitz, H. B., Holden, M. T., Bellonti, J. A., & Ghaffer, A., eds.), pp. 229 –258. Academic Press, New York, NY. 29. Chomczynski, P. & Sacchi, N. (1987) Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal. Biochem. 162: 156 –159. 30. Han, O., Failla, M. L. & Smith, J. C. (1997) Transferrin-iron and proinflammatory cytokines influence iron status and apical iron transport efficiency of Caco-2 intestinal cell line. J. Nutr. Biochem. 8: 585–591. 31. Jung, H. C., Eckmann, L., Yang, S. K., Panja, A., Fierer, J., MorzyckaWroblewska, E., & Kagnoff, M. F. (1995) A distinct array of proinflammatory cytokines is expressed in human colon epithelial cells in response to bacterial invasion. J. Clin. Invest. 95: 55– 65. 32. Glahn, R. P., Lee, O. A., Yeung, A., Goldman, M. I., & Miller, D. D. (1998) Caco-2 cell ferritin formation predicts nonradiolabeled food iron availability in an in vitro digestion/Caco-2 cell culture model. J. Nutr. 128: 1555–1561. 33. Wahl, S. M. (1992) Transforming growth factor beta (TGF-beta) in inflammation: a cause and a cure. J. Clin. Immunol. 12: 61–74. 34. Eckmann, L., Jung, H. C., Schurer-Maly, C., Panja, A., Morzycka-Wroblewska, E., & Kagnoff, M. F. (1993) Differential cytokine expression by human intestinal epithelial cell lines: regulated expression of interleukin 8. Gastroenterology 105: 1689 –1697. 35. Koyama, S. Y. & Podolsky, D. K. (1989) Differential expression of transforming growth factors alpha and beta in rat intestinal epithelial cells. J. Clin. Invest. 83: 1768 –1773. 36. Stadnyk, A. W., Sisson, G. R., & Waterhouse, C. C. (1995) IL-1 alpha is constitutively expressed in the rat intestinal epithelial cell line IEC-6. Exp. Cell Res. 220: 298 –303. 37. Panja, A., Siden, E., & Mayer, L. (1995) Synthesis and regulation of accessory/proinflammatory cytokines by intestinal epithelial cells. Clin. Exp. Immunol. 100: 298 –305. 38. Weiss, G., Fuchs, D., Hausen, A., Reibnegger, G., Werner, E. R., WernerFelmayer, G., & Wachter, H. (1992) Iron modulates interferon-gamma effects in the human myelomonocytic cell line THP-1. Exp. Hematol. 20: 605– 610. 39. Gordeuk, V. R., Prithviraj, P., Dolinar, T., & Brittenham, G. M. (1988) Interleukin 1 administration in mice produces hypoferremia despite neutropenia. J. Clin. Invest 82: 1934 –1938. 40. Philippe, C., Fouqueray, B., Perez, J., & Baud, L. (1992) Up-regulation of tumour necrosis factor-alpha receptors on monocytes by desferrioxamine. Clin. Exp. Immunol. 87: 499 –503. 41. Silver, B. J., Hamilton, B. D., & Toossi, Z. (1997) Suppression of TNF-alpha gene expression by hemin: implications for the role of iron homeostasis in host inflammatory responses. J. Leukoc. Biol. 62: 547–552. 42. Takeuchi, A. (1967) Electron microscope studies of experimental Salmonella infection. I. Penetration into the intestinal epithelium by Salmonella typhimurium. Am. J. Pathol. 50: 109 –136. 43. Giannella, R. A., Formal, S. B., Dammin, G. J., & Collins, H. (1973) Pathogenesis of salmonellosis: studies of fluid secretion, mucosal invasion, and morphologic reaction in the rabbit ileum. J. Clin. Invest. 52: 441– 453. 44. Finlay, B. B. & Falkow, S. (1990) Salmonella interactions with polarized human intestinal Caco-2 epithelial cells. J. Infect. Dis. 162: 1096 –1106. 45. Panigrahi, P., Tall, B. D., Russell, R. G., Detolla, L. J., & Morris, J. G., Jr. (1990) Development of an in vitro model for study of non-O1 Vibrio cholerae virulence using Caco-2 cells. Infect. Immunol. 58: 3415–3424. 46. Francis, C. L., Jerse, A. E., Kaper, J. B., & Falkow, S. (1991) Characterization of interactions of enteropathogenic Escherichia coli O127:H6 with mammalian cells in vitro. J. Infect. Dis. 164: 693–703. 47. Kerneis, S., Chauviere, G., Darfeuille-Michaud, A., Aubel, D., Coconnier, M. H., Joly, B., & Servin, A. L. (1992) Expression of receptors for enterotoxigenic Escherichia coli during enterocytic differentiation of human polarized intestinal epithelial cells in culture. Infect. Immunol. 60: 2572–2580.