Gastrointestinal infections are still a major health problem, not only in developing countries(Reference Schlundt1, Reference Todd2). The enteric pathogen Salmonella enteritidis is one of the leading causes of gastrointestinal infections in humans, ranging from mild, self-limiting diarrhoea and/or inflammation of the intestinal mucosa to life-threatening systemic infection(Reference Mead, Slutsker and Dietz3, 4). The growing resistance of pathogenic bacteria to antibiotic drugs(4) makes it important to develop ways to prevent and treat intestinal infections by other means(Reference Osterholm5). Besides hygienic measures, dietary modulation of host resistance to intestinal infections might be an attractive approach. By influencing the composition of gastrointestinal contents, diet affects the gastrointestinal survival of pathogens(Reference Bovee-Oudenhoven, ten Bruggencate and Lettink-Wissink6, Reference Sprong, Hulstein and Van der Meer7), the composition of autochthonous intestinal microbiota(Reference Bovee-Oudenhoven, Wissink and Wouters8, Reference Kleessen, Sykura and Zunft9) and the epithelial barrier function(Reference Shoda, Mahalanabis and Wahed10). These primary non-immunological host defence mechanisms of the gastrointestinal tract are particularly important in withstanding the first encounter with a pathogen.
Dietary Ca has been recognised to increase resistance by decreasing colonisation and translocation of common intestinal Gram-negative pathogens, both in rats(Reference Bovee-Oudenhoven, Wissink and Wouters8, Reference Bovee-Oudenhoven, Termont and Weerkamp11) and in humans(Reference Bovee-Oudenhoven, Lettink-Wissink and Van Doesburg12). We hypothesise that this resistance-enhancing effect of calcium phosphate might be explained by three potential mechanisms. First, it is known that intake of Ca and phosphate results in the formation of an amorphous calcium phosphate complex that adsorbs and precipitates luminal cytotoxic components, such as bile acids and fatty acids. This can subsequently stimulate growth of protective members of the endogenous microbiota(Reference Bovee-Oudenhoven, Wissink and Wouters8), which exerts antagonistic activity towards foodborne pathogens. Second, Salmonella is a pathogenic micro-organism that can translocate to extra-intestinal organs such as the liver and spleen. Amorphous calcium phosphate-induced precipitation of cytotoxic components reduces epithelial cell damage(Reference Govers, Termont and Van der Meer13) which might strengthen gut barrier function and reduce translocation of Salmonella. Finally, amorphous calcium phosphate might bind Salmonella in the intestinal lumen, thereby preventing Salmonella adherence to the gut mucosa.
Until now, dietary intervention studies have focused on calcium phosphate. From the above-mentioned studies, it can be argued that the combination and luminal interaction of dietary Ca with phosphate is necessary for the resistance-enhancing effect as the adsorbing (binding) capacity of amorphous calcium phosphate is much larger than that of other Ca salts(Reference Govers, Termont and Van Aken14). Little is known about the in vivo effects of other Ca salts on resistance to Salmonella. Considering the high phosphate content of human(Reference Calvo15) and rodent(Reference Reeves, Nielsen and Fahey16) diets, the amorphous calcium phosphate could also be formed with other dietary Ca salts. Therefore, we tested the efficacy of several dietary Ca salts to increase the resistance of rats to Salmonella infection. In addition, in vitro experiments on pH-dependent solubility of the various Ca salts and their capacity to bind Salmonella were determined to gain further mechanistic insight.
Materials and methods
In vitro experiments
Solubility of calcium salts
Only ionic Ca can bind to phosphate and form the amorphous calcium phosphate complex in the gut lumen. Therefore, we studied the solubility of the various Ca salts in the gut lumen. Solubility of the following Ca salts was investigated: calcium phosphate (CaHPO4; purity>98 %; Merck, Darmstadt, Germany), whey product rich in milk Ca (29 % Ca, other minerals and milk salts 66 %; Vivinal® MCA FrieslandCampina Domo, Zwolle, The Netherlands) further referred to as milk Ca, calcium carbonate (purity>99 %; Calcitec V/40S, Mineraria, Italy) or calcium chloride (purity>99 %; Merck). These were tested at a final Ca concentration of 40 mm and at different pH levels representing the pH range of the gastrointestinal tract.The different pH levels ranging from 2 to 7 were achieved by either adding a glycine buffer (250 mm, pH 2 and 3), an acetate buffer (250 mm, pH 4 and 5) or a 3-(N-morpholino) propanesulphonic acid buffer (250 mm, pH 6 and 7). After 1 h incubation at 37°C, samples were centrifuged at 18 000 g for 5 min and supernatants were diluted in 0·5 g/l CsCl. Ca concentrations were determined in the diluted supernatants by inductive coupled plasma-atomic emission spectrophotometry (Varian, Mulgrave, VIC, Australia).
Binding of Salmonella
In order to test whether the amorphous calcium phosphate complex is able to precipitate Salmonella in the gut lumen (in contrast to crystalline calcium phosphate or ionic Ca), we adapted a method that has been previously described to study precipitation of bile salts(Reference Govers, Termont and Van Aken14). Increasing concentrations (final 0–40 mm) of amorphous calcium phosphate, crystalline calcium phosphate or ionic Ca in HEPES buffer (500 mm) were added to a Salmonella suspension in saline (109 colony-forming units/ml). Amorphous calcium phosphate was freshly formed as described elsewhere(Reference Govers, Termont and Van Aken14). In short, equimolar amounts of dissolved calcium chloride and sodium hydrogen phosphate were mixed to form the amorphous calcium phosphate complex. After 15 min incubation at 37°C, samples were centrifuged for 2 min at 500 g. The supernatants were sonicated for 1 min using the Sonicator Ultrasonic Processor XL (Heat Systems, Farmingdale, NY, USA) to disintegrate bacterial cell walls. Total protein, a sensitive way to quantify bacterial content, was determined using the Bradford protein assay (Interchim, Montlucon, France).
In vivo experiment
Animals and diet
The experimental protocol was approved by the animal welfare committee of Wageningen University (Wageningen, The Netherlands). Specific pathogen-free, male, young adult (8 weeks old) Wistar rats with a mean body weight of 276 g (WU, Harlan, Horst, The Netherlands) were housed individually in metabolism cages. All the rats were kept in a temperature- (22–24°C) and humidity-controlled environment (50–60 %) with a 12 h light–dark cycle. To study the effects of the various dietary Ca salts on the main effect variables intestinal permeability and Salmonella persistence, the rats were randomly assigned to either an experimental diet with a total of 20 mmol/kg calcium phosphate (low-Ca negative-control group) or diets supplemented to 100 mmol Ca/kg with calcium phosphate (CaHPO4; purity>98 %; Merck), milk Ca (a whey-derived product; 29 % Ca, other minerals and milk salts 66 %; Vivinal® MCA FrieslandCampina Domo), calcium carbonate (purity>99 %; Calcitec V/40S, Mineraria) or calcium chloride (purity>99 %; Merck). Diets and demineralised drinking water were supplied ad libitum. Compared with the AIN-93 recommendation for rat diets(Reference Reeves, Nielsen and Fahey16), the diets had a high fat content (200 g fat/kg provided by 80 g/kg maize oil and 120 g/kg palm oil) to mimic the composition of a Western human diet. The composition of the diets has been described earlier(Reference Ten Bruggencate, Bovee-Oudenhoven and Lettink-Wissink17). Food intake and body weight were measured every 2–3 d before infection and daily after infection. After infection, the average of daily food intake (kJ/d) and growth was calculated per animal.
Calcium excretion
Ca was determined in freeze-dried faeces after dry-ashing and destruction. Faeces were treated with 50 g/l of TCA (1:1, v/v) and centrifuged for 2 min at 14 000 g. The supernatants were diluted with 0·5 g/l CsCl and analysed by inductive coupled plasma-atomic emission spectrophotometry (Varian).
Urine was treated with 50 g/l of TCA (1:1, v/v) and centrifuged for 2 min at 14 000 g. The supernatants were analysed as described for faeces. Since daily feed intake differed between animals and between dietary groups, faecal and urinary Ca excretion was calculated as percentage of intake.
Composition of the intestinal microbiota
Fresh faecal samples collected 1 d before infection were analysed for the number of lactobacilli and enterobacteria. Faecal lactobacilli were quantified by plating appropriate 10-fold dilutions on Rogosa agar (Oxoid, prepared according to instructions of the manufacturer) and incubating the plates in an anaerobic cabinet (Coy Laboratory Products, Inc., Ann Arbor, MI, USA) under an anaerobic gas mixture (85 % N2, 10 % CO2 and 5 % H2) at 37°C for 3 d as described previously(Reference Giaffer, Holdsworth and Duerden18). Enterobacteria were determined by plating 10-fold dilutions on Levine EMB agar (Difco Laboratories, Detroit, MI, USA) and incubating overnight at 37°C(Reference Giaffer, Holdsworth and Duerden18).
Intestinal permeability
In order to measure intestinal permeability, the marker Cr-EDTA was added to all the diets. Cr-EDTA, an inert complex that is not actively taken up by the mucosa(Reference Bjarnason, Maxton and Reynolds19, Reference Oman, Blomquist and Henriksson20), was prepared as described elsewhere(Reference Binnerts, Van het Klooster and Frens21). Total 24 h urine samples were collected on the last day before and on 7 consecutive days after oral infection of the rats. Oxytetracycline (1 mg) was added to the urine collection vessels of the metabolic cages daily to prevent bacterial deterioration. To measure Cr-EDTA, urine was treated with 50 g/l of TCA (1:1, v/v) and centrifuged for 2 min at 14 000 g. The supernatants were diluted with 0·5 g/l CsCl and Cr was analysed by inductive coupled plasma-atomic emission spectrophotometry. Urinary Cr-EDTA excretion was calculated as a percentage of daily dietary Cr-EDTA intake.
Oral infection with Salmonella
After adaptation to the housing and dietary conditions for 2 weeks, the rats were orally infected by gastric gavage of 109 colony-forming units of S. enteritidis (clinical isolate, phage type 4 according to international standards; B1241 culture of NIZO food research, Ede, the Netherlands) suspended in 1 ml of saline. Stability of the infection stock solution was determined by plating serial dilutions on Modified Brilliant Green Agar (Oxoid) containing sulphmandelate (Oxoid) immediately before and after the oral dosing. Salmonella was cultured and stored as described earlier(Reference Bovee-Oudenhoven, Termont and Heidt22). On day 10 after oral infection, the rats were killed by CO2 inhalation.
Persistence of Salmonella
Immediately before and on days 1, 2, 5 and 7 after S. enteritidis infection, fresh faecal samples were collected directly from the anus of the animals and analysed for viable Salmonella by plating 10-fold dilutions on Modified Brilliant Green Agar (Oxoid) containing sulphmandelate (Oxoid) and incubating overnight at 37°C.
Statistical analysis
Results are expressed as means with their standard errors, n 7 in the calcium phosphate group and n 8 in the other groups. One animal from the calcium phosphate group was excluded from all the results because of oral–pharyngeal reflux of the Salmonella suspension resulting in pneumonia. A commercially available package (Statistica 6.1; StatSoft, Inc., Tulsa, OK, USA) was used for all statistics.
All dietary Ca groups were only compared to the low-Ca control group. In case of normally distributed data (as indicated by the Shapiro–Wilk test), differences between means were tested for their significance using a one-way ANOVA, followed by the Student t test (two sided). When data were not normally distributed, differences were tested for their significance using a Kruskall–Wallis ANOVA, followed by the non-parametric Mann–Whitney U test (two sided). Repeated-measures ANOVA was used for intestinal permeability and Salmonella colonisation. Bonferroni correction was used for multiple testing (four comparisons). Differences were considered statistically significant when P < 0·05.
Results
In vitro experiments
Solubility of calcium salts and binding of Salmonella
Solubility of the different Ca salts was determined at pH 2–7 (Fig. 1), representing the pH range in the gastrointestinal tract. All the Ca salts were soluble at lower (gastric) pH. Calcium chloride was completely soluble at all pH levels tested. The other Ca salts were most soluble at pH 5–7.
Fig. 1 Solubility of different Ca salts as a function of pH. Values are exemplary for a few studies with different Ca salts. ○, Calcium phosphate; □, milk Ca; ⋄, Ca carbonate; △, calcium chloride.
Freshly formed amorphous calcium phosphate was capable of binding Salmonella. In contrast, crystalline calcium phosphate and ionic Ca showed almost no binding capacity (Fig. 2).
Fig. 2 Percentage of Salmonella bound to amorphous or crystalline Calcium phosphate or ionic Ca. Results are expressed as means and standard deviations. Values are means of triplicate incubations. Standard deviations are either shown or smaller than symbols. ■, Amorphous calcium phosphate; ♦, ionic Ca; ▲, crystalline calcium phosphate.
In vivo experiment
Calcium excretion and composition of the intestinal microbiota
Before infection, faecal Ca excretion (as a percentage of total intake) was higher in all Ca-supplemented groups compared to the low-Ca control group. Urinary Ca excretion was lower in the calcium phosphate and milk Ca group compared to the low-Ca control group (Fig. 3).
Fig. 3 Faecal and urinary Ca excretion (% of intake) before infection of the rats. Ca was determined by inductively coupled plasma-atomic emission spectrophotometry. Results are expressed as means with their standard errors (n 7 in the calcium phosphate group and n 8 in the other groups). ■, Low-Ca control; □, calcium phosphate; ▨, milk Ca; 
, calcium carbonate; 
, calcium chloride. * The indicated group is significantly different from the low-Ca control group (P < 0·05).
Before infection with Salmonella, no differences in faecal lactobacilli were observed between the diet groups. Calcium phosphate and milk Ca reduced the number of enterobacteria compared to the low-Ca control group (Fig. 4).
Fig. 4 Effect of dietary Ca supplementation on enterobacteria and lactobacilli in faecal samples collected before Salmonella infection of the rats. Lactobacilli were cultured anaerobically on Rogosa agar and enterobacteria were cultured aerobically on Levine EMB agar. Results are expressed as means with their standard errors (n 7 in the calcium phosphate group and n 8 in the other groups). ■, Low-Ca control; □, calcium phosphate; ▨, milk Ca; , calcium carbonate; , calcium chloride. * The indicated group is significantly different from the low-Ca control group (P < 0·05). CFU, colony-forming units.
Food intake and body weight
Food consumption before infection was approximately 22 g and did not differ between the diet groups. During the first week after infection, mean food intake was significantly higher in the calcium phosphate group compared to the low-Ca group (Fig. 5(a)).
Fig. 5 Effect of dietary Ca supplementation on mean food intake (a) and growth of the rats (b) after oral administration of 1 × 109 colony-forming units of Salmonella enteritidis. Results are expressed as means (n 7 in the calcium phosphate group and n 8 in the other groups). ■, Low-Ca control; □, calcium phosphate; ▨, milk Ca; , calcium carbonate; , calcium chloride. * The indicated group is significantly different from the low-Ca control group (P < 0·05).
Body weight gain before infection was similar in all the diet groups. However, after infection average body weight gain was significantly higher in the calcium phosphate-supplemented group (Fig. 5(b)).
Intestinal permeability
Due to ad libitum feeding of the animals, daily dietary intake of Cr-EDTA (marker for intestinal permeability) differed between days and animals. Therefore, urinary Cr-EDTA excretion was calculated as a percentage of total daily dietary intakes. Urinary Cr-EDTA in the low-Ca control group strongly increased after infection, reaching a maximum at day 5 (Fig. 6). Infection-induced intestinal permeability was lower in all Ca-supplemented groups compared to the low-Ca control group. However, the kinetics of urinary Cr-EDTA was similar in all the diet groups and illustrate Salmonella-induced damage to the intestinal barrier.
Fig. 6 Effects of dietary Ca supplementation on urinary Cr-EDTA excretion in rats. Daily urinary Cr-EDTA excretion is expressed as % of dietary intake. Rats were challenged with 1 × 109 colony-forming units of Salmonella enteritidis on day 0. Cr-EDTA was analysed by inductively coupled plasma-atomic emission spectrophotometry. Results are expressed as means with their standard errors (n 7 in the calcium phosphate group and n 8 in the other groups). ●, Low-Ca control; ○, calcium phosphate; □, milk Ca; ⋄, calcium carbonate; △, calcium chloride. All the Ca-supplemented groups were significantly different from the low-Ca control group (P < 0·05).
Persistence of Salmonella
All the Ca-supplemented groups decreased persistence of Salmonella, as indicated by decreased faecal excretion of Salmonella with time (Fig. 7). In contrast, rats fed the low-Ca diet continued to excrete high Salmonella numbers in their faeces.
Fig. 7 Effect of dietary Ca supplementation on faecal Salmonella excretion with time. Rats orally received 1 × 109 colony-forming units (CFU) of Salmonella enteritidis on day 0. Results are expressed as means with their standard errors (n 7 in the calcium phosphate group and n 8 in the other groups). ●, Low-Ca control; ○, calcium phosphate; □, milk Ca; ⋄, calcium carbonate; △, calcium chloride. All the Ca-supplemented groups were significantly different from the low-Ca control group (P < 0·05).
Discussion
The present study shows that besides calcium phosphate, also milk Ca, calcium carbonate and calcium chloride are able to enhance the resistance to infection, as indicated by the reduction of infection-induced intestinal permeability and the decreased persistence of Salmonella in time. The calcium phosphate-induced improvement of intestinal resistance to Salmonella is in agreement with earlier studies(Reference Bovee-Oudenhoven, Lettink-Wissink and Van Doesburg12, Reference Ten Bruggencate, Bovee-Oudenhoven and Lettink-Wissink23, Reference van Ampting, Rodenburg and Vink24).
Protection against Salmonella by direct binding of Salmonella to the amorphous calcium phosphate complex seems a likely explanation for the observed protective effect. The amorphous calcium phosphate complex is formed in the human(Reference Govers, Termont and Lapre25, Reference Van der Meer, Welberg and Kuipers26) and rat(Reference Bovee-Oudenhoven, Wissink and Wouters27) proximal small intestine from soluble ionic Ca together with phosphate originating from the diet. The amorphous calcium phosphate complex could be formed in vivo with all tested Ca salts; soluble ionic Ca was present after gastric passage in vitro, Ca was abundantly present in the intestinal lumen (as indicated by the faecal Ca excretion), and all the diets contained phosphate (originating from acid casein). The in vitro experiments also showed that amorphous calcium phosphate, in contrast to crystalline calcium phosphate and ionic Ca, was able to bind Salmonella. Binding of Salmonella to the amorphous calcium phosphate complex in the intestinal lumen may subsequently prevent mucosal attachment of these pathogens and inhibit infection-induced pathology. This interaction may also play a role in the observed dietary Ca-induced reduction in faecal enterobacteria in the present study.
Besides direct binding of Salmonella during gut transit, amorphous calcium phosphate can also precipitate cytotoxic luminal components, such as bile acids and fatty acids, within the intestinal lumen(Reference Govers, Termont and Lapre25–Reference Bovee-Oudenhoven, Wissink and Wouters27). Binding of these surfactants can subsequently prevent epithelial cell damage and compensatory epithelial hyperproliferation(Reference Van der Meer, Welberg and Kuipers26, Reference Sesink, Termont and Kleibeuker28), and may protect the intestinal barrier(Reference Govers, Termont and Lapre25, Reference Van der Meer, Welberg and Kuipers26, Reference Govers and Van der Meer29). In fact, in the present study, all the tested dietary Ca salts prevented the infection-related increase in intestinal permeability.
The protective effects of dietary Ca are not restricted to Salmonella. We have previously shown that dairy Ca strongly improves the resistance to enterotoxigenic Escherichia coli in both rats and human subjects. To follow-up, a large intervention trial is currently being performed to determine whether milk Ca can protect against acute infectious diarrhoea in Indonesian children.
Are the dietary concentrations of Ca in our animal studies relevant for the human diet? In general, dietary Ca intake in the Western world ranges from 600 to 1100 mg/d(Reference Buchowski, Semenya and Johnson30, Reference Lovejoy, Champagne and Smith31). Assuming a daily food intake of 500 g dry weight per d, the average concentration in the food would be 1·2–2·2 g/kg. The animal diets in this mechanistic study contained Ca levels of 0·8 g/kg (20 mmol/kg diet) and 4·0 g/kg (diets supplemented to 100 mmol/kg), and therefore represent conditions slightly lower and higher than normal human intake.
Overall, the present study showed that several dietary Ca salts (calcium phosphate, milk Ca, calcium carbonate, calcium chloride) are protective in improving resistance to Salmonella infection. This protective effect might be explained by binding of Salmonella (thereby preventing mucosal attachment of Salmonella) and/or binding of cytotoxic luminal components (thereby protecting the intestinal barrier). In view of the growing resistance of pathogenic bacteria to antibiotic drugs, modulation of host resistance to intestinal infections by dietary Ca might be an attractive approach.
Acknowledgements
The authors thank the workers at the laboratory animal facility of Wageningen University for technical assistance. M. H. C. S. and E. E. are employees of FrieslandCampina Domo (Zwolle, The Netherlands) that is a producer of milk Ca used in the present study. The remaining authors declare no conflict of interest. No external funding, apart from the authors' institutions, was available for the present study. All the authors contributed to the preparation of the paper and agreed with the submitted manuscript content. S. J. M. t. B., J. S., M. H. C. S. and E. E. designed the research; E. v. d. M. and A. S. performed the research and analysed the data; and. S. J. M. t. B., J. S. and I. M. J. B.-O. drafted the paper.
