Main

Iron (Fe) deficiency is one of the most common nutritional problems around the world, both in developed and in developing countries. It is responsible for a long-lasting impairment of neurologic development in infants; Fe is also essential for immunity and growth (1,2).

Cow milk products and cow milk–based infant formulas are two main components of infant diet; however, Fe absorption from cow milk is low compared with breast milk (3); the high calcium concentration and the kind of proteins of cow milk could explain these differences (4,5). The main cow milk proteins are caseins, which bind cations, including Fe, by clusters of phosphoserine (6,7). This strong binding keeps Fe soluble at the alkaline gut pH but prevents its release in a free form available for absorption by the duodenal mucosa (8). Hydrolysis of whole caseins improves Fe absorption (5), but caseinophosphopeptides (CPP) released by the enzyme digestion of different caseins give conflicting results; the two main cow milk caseins are αS- and β-caseins; binding Fe to CPP of β-casein or to hydrolyzed β-casein improves its absorption and its bioavailability in rats and in humans (9–11); however, CPP derived from the hydrolysis of whole casein or fractions enriched with αS-casein CPP inhibit Fe absorption (12–14).

One of the key steps of the absorption of CPP-bound Fe seems to be the release of phosphoserine residues by brush border phosphatase: it could be assumed that releasing free Fe could increase its absorption from a poorly absorbed form; however, releasing Fe from a stable, efficiently absorbed complex, such as Fe-βCPP (1–25) could impair its absorption by increasing the amount of free Fe subject to gut interactions. These functional differences between caseins and casein-derived CPP could give some clues to improve the bioavailability of cow milk Fe.

The objectives of this study were to evaluate the absorption of Fe complexed to CPP purified either from β-casein or from αs1-casein and to determine the effect of an inhibitor of intestinal alkaline phosphatase (Na2WO4) (12,15). The study used two models, the in vivo perfused rat intestinal loop and a human colon adenocarcinoma cell line culture (Caco-2), which is predictive of human Fe bioavailability (16).

METHODS

Preparation of phosphopeptides and phosphopeptide/Fe complexes.

Pure β-casein and αs-caseins (a mixture of αs1- and αs2-caseins) were first purified from whole bovine caseins by Fast Protein Liquid chromatography on an anion exchange column according to the method described by Guillou et al. (17). Purified β-casein and αs-caseins then were digested by trypsin to isolate corresponding phosphorylated sequences as previously reported (9). The phosphopeptides were characterized by electrospray mass spectrometry working on-line with an HPLC (14). Single, pure phosphopeptide [β-CPP (1–25)] was obtained from β-casein, whereas a mixture that contained αs1-CPP (59–79) and αs2-CPP (2–21) was obtained from αs-casein digest. The characteristics of the CPP preparations are summarized in Table 1.

Table 1 Characteristics of purified caseinophosphopeptides (CPP)
Full size table

Binding of Fe to the purified CPP was performed by mixing them with FeCl2 solution (Fe/phosphopeptides molar ratio = 4) for 1 h at 37°C. Free Fe was removed by diafiltration of the solutions on a 3-kD ultrafiltration membrane. The amount of Fe bound to the CPP was determined by atomic absorption spectrometry (Varian, Model AA 1275; Les Ulis, France) on freeze-dried samples. Double-distilled and deionized water was used. All glassware and the polyethylene tubes used for the samples were washed and rinsed in distilled water.

In vitro model: Caco-2 cell culture.

The human colon adenocarcinoma cell line Caco-2 was obtained from the American Type Tissue Collection and studied between the 40th and 45th passages. All cell culture solutions were obtained from Life Technologies (GIBCO, Paisley, Scotland, NY). Caco-2 cells were routinely cultured in DMEM supplemented with 10% heat-inactivated FCS, 4 mM l-glutamine, 1% antibiotics mixture (100 U/mL penicillin and 100 μg/mL streptomycin), and 1% nonessential amino acids. They were cultured in a 75-cm2 tissue culture flask (Falcon, Franklin Lakes, NJ) at 37°C under a 5% CO2/95% air atmosphere. When confluent, cells were dispersed by treatment with Earle's salt solution that contained 0.05% trypsin and 0.53 mM EDTA and seeded into 12-well cell culture at a density of 250,000 cells per insert (Falcon). The medium was changed every 2–3 d.

Tightness of the cell monolayer grown on filters.

Caco-2 monolayers were grown for 21 d; they were used for transport studies when the monolayer was intact, as checked by the transepithelial resistance measured with an epithelial volt-ohmmeter with dual electrodes (Word precision Instruments, New Haven, CT). Filters were used only when the electrical resistance of monolayers exceeded 250 Ω/cm2

Transport studies by Caco-2 cells.

Transepithelial transport was studied with cells that were grown on filters for 21 d. Transport studies used a special DMEM (GIBCO) without Fe and without phenol red (spDMEM). Monolayers were rinsed twice with spDMEM at 37°C and transferred to fresh 12-well culture dishes that contained 1.5 mL of Fe-free transport medium 30 min before the beginning of the experiment. Three Fe transport solutions that contained 100 μM Fe (Fe gluconate as control, Fe–β-CPP, Fe–αS1-CPP) were formulated. For achieving intestinal alkaline phosphatase activity, 1 mmol/L Na2WO4 was added to the three forms of Fe uptake medium (Fe gluconate + Na2WO4, Fe–β-CPP + Na2WO4, Fe–αS1-CPP + Na2WO4). Then, 1 mL of Fe solutions was added to the top compartment of the insert, and plates were placed on an incubator (37°C, low oscillations) for 2 h. After incubation, uptake into cell monolayers was stopped by addition of ice-cold spDMEM, and the transepithelial resistance was measured. Solutions were removed by aspiration, and monolayers were incubated once with 500 μmol of 1 mM bathophenanthroline disulfonic acid and 5 mM sodium hydrosulfite solution to remove surface-bound Fe (12). Cell monolayers then were rinsed twice with Hank's buffer solution salt without Ca and Mg and solubilized with 0.1% triton (2 mL).

For all groups, six repetitions were made, and for Fe assay, samples that contained aliquots of cell cultures were assayed on an atomic absorption spectrometer (Perkin-Elmer Perkin Elmer F-91945 Courtaboeuf, France). Results were expressed in nmol Fe/2 h.

In vivo model: perfused rat intestinal loop.

Adult Sprague-Dawley rats that weighed 200–250 g were starved for 12 h before study but had free access to water. Six were perfused with one of the three forms (Fe Gluconate, Fe–αs1-CPP or Fe–β-CPP); experiments were performed without (controls) or with the addition of Na2WO4. The composition of perfusion solute was adapted from Ringer-Lavoisier solute: its pH was adjusted to 7 (duodenum pH); it was isotonic to plasma (285–300 mOsmol/L) and contained 100 μM Fe. When necessary, Na2WO4 was added at a concentration of 1 mM.

Anesthesia was performed with ketamine (Ketalar, Pfizer Laboratories, Paris, France). Duodenum was exposed by a laparotomy; it was perfused through a catheter inserted into the pylorus, and effluent was collected at the angle of Treitz. Every element of the perfusion device was previously washed with a solution of Triton-X 100 (1 g/L) to prevent any contamination and to cleave Fe, which could be adsorbed to the mucosa. The perfusion solute (20 mL) was kept at 37°C by thermostatic control and was delivered at 0.16 mL/min, using a peristaltic pump, to avoid loop distention. After 2 h of perfusion, the animal was killed by an overdose of Dolethal (10). Fe concentration was measured by atomic spectrometric absorption (Perkin-Elmer 3030) on perfusion solute, digestive effluent, and mucosa of the perfused segment; the minimal detection level of the method is 0.19 μmol Fe/L.

Statistical analysis.

Results of Fe absorption were expressed as means and SD. Groups were compared by two-way ANOVA and t test. Differences were considered significant at p < 0.05. Data analyses were performed using the Stat View SE+Graphics statistical program.

RESULTS

Results of Fe absorption are given in Table 2. Total uptake of Fe from Fe–β-CPP complex was higher than Fe gluconate in both experimental models. Net absorption of Fe gluconate by the Caco-2 cell was significantly higher than Fe–β-CPP. No significant difference was displayed between the net absorption of Fe gluconate and of Fe–β-CPP complex by the duodenal loop. Fe uptake and net absorption from Fe–αS1-CPP complex were significantly lower than from the other forms of Fe, in both models. Indeed, in contrast to the two other forms of Fe, high mucosal storage of Fe was observed with Fe–αS1-CPP complex in the duodenal loop system.

Table 2 Effect of caseinophosphopeptide forms on iron uptake and absorption. Initial concentration in upper compartment of cell culture and in perfusion solute: 100 μmol/l. Total amount in nmol after 2 hours of experiment. N = 6/group
Full size table

The dependence of Fe absorption on the origin and the sequence of CPP is further illustrated in Fig. 1, which shows the level of Fe absorption from Fe–αs1-CPP compared with Fe–β-CPP. Fe absorption from β-CPP (1–25) complex was at least two times higher than absorption from Fe–αS1-CPP complex in the duodenal loop model. This difference was more pronounced with Caco-2 cells.

Figure 1
figure 1

Fe absorption by Caco-2 cells and in vivo–perfused rat duodenal loop as affected by Fe carriers. The following forms of Fe were compared: Fe gluconate, Fe–αS1-CPP, and Fe–β-CPP (initial Fe concentration: 100 μM). n = 6/group; *different from non inhibited group: p < 0.05; **different from non inhibited group: p < 0.01.

Full size image

The effects of inhibition of intestinal phosphatase on Fe uptake and absorption by Caco-2 cells and perfused duodenal loop are displayed in Fig. 2. No significant effect of the inhibitor was observed for Fe gluconate or for Fe–αS1-CPP. In contrast, the inhibition of phosphatase significantly increased the uptake and the absorption of Fe from Fe–β-CPP complex.

Figure 2
figure 2

Inhibition (% of control) of Fe absorption from the different forms by a specific phosphatase inhibitor (1 mM Na2WO4) as assessed with Caco-2 cells model (A) and perfused rat duodenal loop model (B). Initial Fe concentration: 100 μM; pH 7. Values are means ± SD; n = 6/group; *different from noninhibited group: p < 0.001; **different from noninhibited group: p < 0.0001.

Full size image

DISCUSSION

Cow milk and cow milk products are an essential part of infant diet, yet they present a poor Fe absorption and so could contribute to the high prevalence of Fe deficiency in infants around the world; the affinity of cow milk proteins for Fe is so strong that it cannot be released and absorbed in a free form. Hydrolysis of cow milk proteins enhances Fe absorption (5), but previous studies suggested that not all caseins are equal in their interactions with Fe. Whereas CPPs derived from β-casein enhance Fe absorption and bioavailability in experimental and human studies (9–11), CPPs derived from whole caseins or from αs1-casein–enriched fractions seem to exert an opposite effect (12–14). The confirmation of such functional differences between milk proteins could have significant consequences on infant nutrition, because breast milk is rich in β-casein but has no α-casein.

The ability of the two CPP preparations [β-CPP (1–25) and αs-CPP (containing mainly the sequence αs1-CPP [59–79])] to improve Fe absorption was compared. These two phosphopeptides contain four and five phosphorylated serine residues, respectively. Their amino acid sequences are shown below using the one-letter code, with the Ser(P) (Sp) cluster sequence in italics:

αs1-CPP (59–79): 59QMEAESpISpSpSpEEIVPNSpVEQK79

β-CPP (1–25): 1RElEELNVPGEIVESpLSpSpSpEESITR25

The two experimental models that we used, in vivo intestinal perfusion and in vitro Caco-2 cell culture, gave the same results; they confirmed the efficient uptake and absorption of Fe bound to the specific CPP of β-casein [β-CPP (1–25)], which is similar to pharmaceutical forms of Fe (Fe gluconate) and better than the reference salt FeSO4 (data not shown). It should be kept in mind that these results were obtained in basal conditions; a better absorption of β-CPP (1–25) bound Fe compared with gluconate was observed during Fe deficiency (10).

For the first time, purified forms of αs-casein–derived CPP were used and were shown to significantly lower Fe uptake and absorption. Similar observations on the relationships between the structure of CPP and mineral metabolism were recently made by Ferraretto et al. (18). They showed that the CPP-promoting effect on calcium uptake by differentiated HT-29 cells depends on the conformation of the CPP in solution. More general, these results underline that the physico-chemical characteristics of food-derived peptides must be taken into account when assessing their biologic activities.

The main differences between the two CPPs presently used concern Fe uptake, so this study assessed the influence of releasing Fe from CPP complexes in the immediate vicinity of the apical membrane of enterocyte. As expected, inhibiting the brush border phosphatase activity had no significant effect on the metabolism of Fe gluconate; it did not modify the absorption of αS1-CPP (59–79)–bound Fe, probably as a consequence of the low accessibility of the phosphoserie residues of this complex to intestine enzymes as discussed below. The significant enhancement of uptake of Fe–β-CPP (1–25) after inhibition of the dephosphorylation reaction means that free Fe released from this complex is less available than the bound form. Although β-CPP (1–25) and αs1-CPP (59–79) possess a common acidic motif (italic region of their structure), they differently affected the uptake and the absorption of Fe in both intestinal loop and Caco-2 cell models. The observed difference in the biologic function of these peptides may be related to different structure and/or conformation in solution. In fact, recent nuclear magnetic resonance studies have shown that despite their high degree of sequential and functional similarity, β-CPP (1–25) and αs1-CPP (59–79) have distinctly different conformations in solution in the presence of calcium ions (19). In the same way, binding of Fe may have induced different conformations of these molecules. Hence, a specific conformation of αs1-CPP (59–79) in solution combined with its higher electronegative property (five phosphoserine residues) could generate a too tight Fe/peptide complex and consequently a lower bioavailability of the mineral.

This study gives for the first time experimental evidence to explain the conflicting results on the exact role of CPP in mineral bioavailability. Indeed, we believe that the previous results are conflicting only in appearance: cow milk proteins have different functional properties concerning mineral metabolism; β-casein and its CPP enhance Fe absorption, whereas αS-casein–rich fractions or their CPP inhibit it (12,13). The impairing effect of releasing Fe from the Fe–β-CPP complex at the brush border membrane level suggests, as previously shown (10), that its uptake by the mucosa occurs in a bound form.