![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
Department of Preventive Medicine and Community Health, University of Texas Medical Branch, Galveston, TX 77555-1109 and * Department of Food Science and Human Nutrition, Iowa State University, Ames, IA 50011-1120
3To whom correspondence should be addressed. E-mail: reserfas{at}utmb.edu.
 |
ABSTRACT |
|---|
 TOP ABSTRACT MATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
4.2 kDa that contained iron and that solubilized added ferric iron both before and after in vitro digestion with pepsin, pancreatin and bile extract. Our results suggested that a low MW component of breast milk whey enhances iron bioavailability. Because the iron solubilization activity is resistant to in vitro digestion, it is plausible that the component is active in vitro and may explain the excellent absorption of EFe by breastfed infants.
KEY WORDS: • Caco-2 cells • iron absorption • breast milk • inductively coupled plasma mass spectrometry
Breastfed infants absorb extrinsic (nonmilk) iron exceptionally well (1
). Absorption of iron and its incorporation into hemoglobin measured by the extrinsic tag method are greater in breastfed than in formula-fed term infants (2
) and have been correlated inversely to plasma ferritin concentration in breastfed term infants (2
,3
). Absorption of extrinsic iron by fasted adults is greater in the presence of human milk than in the presence of cow’s milk (4
). The higher bioavailability of iron when breast milk is given is not fully explained by the factors known to affect iron absorption (5
). Once it was thought that lactoferrin, abundant in human but not in cow’s milk, might facilitate iron absorption (6
). However, absorption of extrinsic iron has been shown to be slightly greater when lactoferrin is absent than when it is present (7
).
Our objective was to identify components in human milk that enhance extrinsic iron absorption. Sequential processes obligatory to absorption of extrinsic iron include iron solubilization, iron uptake into and passage through the enterocyte and iron transfer across the basolateral membrane to the circulation. Monolayers of Caco-2 cells properly prepared on porous supports facilitate precise comparisons of the effects of treatment variables on these mechanistic processes (8
–15
). Caco-2 cells modify their ferrokinetics with changes in cellular iron status, possess the proper morphology and cellular constituents and exhibit characteristics similar to those observed in human studies (8
,14
,16
,17
). Moreover, they are useful for quantification of food iron bioavailability (17
).
In this research, we used monolayers of Caco-2 cells to screen fractions of breast milk for their ability to enhance transepithelial flux of extrinsic iron. Then some of the most stimulatory fractions were studied for their influence on iron solubilization. Finally, the most stimulatory of these fractions was characterized as to the molecular weight of the active component, its iron binding capacity and its resistance to in vitro digestion. For clarity and brevity, our findings are reported in the sequence that these physiologic processes normally occur rather than in the order the experiments were conducted.
The amount of iron intrinsic to breast milk is insufficient to meet the iron needs of breastfed infants for longer than a few months even if one assumes the intrinsic iron is absorbed completely. Extrinsic, nonmilk iron is present in the gastrointestinal tract of infants from endogenous secretions and from exogenous sources such as supplements and other foods. We hypothesized that persistence of specific peptides from breast milk in the proximal small intestine facilitates absorption of extrinsic iron and contributes beneficially to the iron status of the breastfed infant.
We report here the existence of several breast milk fractions that promote solubilization of extrinsic iron at intestinal pH and enhance flux of extrinsic iron across Caco-2 cell monolayers. At least one of these fractions has substantial iron content, binds an appreciable amount of extrinsic iron and is resistant to digestion by pepsin, pancreatin and bile extract.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Antibiotic-antimycotic solution, nonessential amino acids and trypsin-EDTA (4
) were purchased from GIBCO (Grand Island, NY). Earle’s balanced salt solution without bicarbonate (EBSS),4 fetal bovine serum, HEPES, 2-(N-morpholino)ethanesulfonic acid (MES), SDS, L-ascorbic acid, nitrilotriacetic acid (NTA), Folin-Phenol reagent, Dulbecco’s modified Eagle’s medium, pepsin, pancreatin, bile extract, thioglycolic acid, human 
-lactalbumin, Ferrozine, ultra-low range molecular weight markers and all other reagents not otherwise noted were supplied by Sigma Chemical (St. Louis, MO). American Type Culture Collection (Rockville, MD) supplied Caco-2 cells. Bicameral 6-well collagen-treated culture plates were purchased from Costar (Cambridge, MA). 59FeCl3 was purchased from Du Pont NEN (Boston, MA). Reagent grade FeCl3 · 6H2O, methanol, acetic acid and trace metal grade acids and bases were purchased from Fisher Scientific (Pittsburgh, PA). Plastic labware was purchased from Fisher Scientific or Nalge Nunc International (Rochester, NY). Type I (American Society for Testing Materials) water was supplied by a Nanopure II analytical grade water system (Barnstead/Thermolyne, Dubuque, IA) in a Class 100 clean room (Controlled Environment Structures, Mansfield, MA). Ultrahigh purity argon was supplied by Aeriform (Houston, TX). Polyethyletherketone (PEEK) tubing and fittings and polyvinylchloride tubing were purchased from Cetac Technologies (Omaha, NE).
Fractionation of human milk.
Aliquots of surplus human milk from samples collected under approved research protocols were combined by personnel of the Pediatric Metabolism Unit of the University of Iowa and were stored frozen at -20°C for 1–4 wk before use. Acquisition of samples of human milk in this manner was approved by the Committee on Use of Human Subjects in Research of Iowa State University. Storage was in accordance with reports that proteolysis is minimal under these circumstances (18
). Aliquots of thawed, pooled human milk were fractionated as shown in Figure 1
. Milk samples were subjected to centrifugation at 4,000 x g for 30 min at 4°C, and the upper, fat-containing fraction was removed and saved. The remainder ("Skim Milk") received ultracentrifugation at 150,000 x g for 60 min at 4°C that resulted in a pellet ("Casein") and a supernatant portion ("Whey"). The whey was subjected to ultrafiltration through a 10-kDa molecular weight cutoff membrane to obtain 10-kDa retentate (10KR) and 10-kDa filtrate (10KF). The 10KF was then subjected to ultrafiltration using a 3-kDa molecular weight cutoff membrane to obtain retentate (3KR) and filtrate (3KF) fractions. Subsequently, the 3KF was subjected to ultrafiltration using a 1-kDa molecular weight cutoff membrane to produce retentate (1KR) and filtrate (1KF). An Amicon ultrafiltration system (Model 8200 with YM 10, YM 3 and YM 1 membranes; Amicon, Beverly, MA) was used at 4°C under nitrogen for 16–24 h.
|
Determinations of protein and iron.
Milk and milk fractions were lyophilized in perfluoroacetoxy vials (Savillex, Minneapolis, MN), predigested with trace metal grade nitric acid (70%) overnight at room temperature and digested in the vials within closed vessels in a microwave digestion apparatus (Milestone 1200 Mega; Milestone, Monroe, CT). Iron in fractions was determined by using flame atomic absorption spectrophotometry versus working standard solutions prepared from spectrometric standard iron from the National Institute of Standards and Technology (Gaithersburg, MD). Iron in uptake buffer and in chromatographic eluates of some milk fractions before and after incubations was determined by using inductively coupled plasma mass spectrometry (ICPMS). Protein was determined by the Lowry method (19
).
Preparation of fractions for studies of iron solubilization and transmonolayer flux.
Apical (uptake) buffer was freshly prepared by addition of 10 mmol of MES/L to EBSS to make EBSS/MES at pH 6.0 with or without the addition of phenol red (100 µmol/L) as desired. Iron concentration of freshly prepared uptake buffer was 0.10 ± 0.05 (mean ± SD) µmol/L as determined by ICPMS. Lyophilized aliquots of milk fractions typically were reconstituted to a protein concentration of 2 g/L in EBSS/MES. To simulate exposure to infant gastric fluid, aliquots of each reconstituted fraction usually were brought to pH 3.0 by dropwise addition of 5 mol of HCl/L and warmed for 15 min at 37°C. Then all aliquots usually were brought to pH 6.0 by dropwise addition of 1 mol of NaHCO3/L and maintained at 37°C for 12 min before use in experiments of solubilization and transmonolayer flux of iron. For characterization of solubilization by using ICPMS, the solute limitations of ICPMS were met by omission of EBSS, diminution of the MES concentration from 10 mmol/L to 0.1–0.3 mmol/L and increase in the pH of MES from 6.0 to 6.5–7.0 by the addition of ammonium hydroxide.
Stock solutions contained ascorbic acid (20 mmol/L in H2O), nitrilotriacetic acid (5 mmol/L in H2O), ferrous sulfate (1 mmol/L in 0.1 mmol of HCl/L), ferric chloride (1 mmol/L in 10 mmol of HCl/L) and 59FeCl3 (specific radioactivity, 32.3 GBq/mmol; 0.573 µmol of Fe/L in 10 mmol of HCl/L). For studies of transmonolayer iron flux, ferrous/ascorbate control solutions (10 and 15 µmol of iron/L, ferrous/ascorbate molar ratio 1:20) were prepared freshly by the addition of 1 or 1.5% vol/vol ascorbic acid and ferrous sulfate stock solutions to EBSS/MES, pH 6.0.
Iron solubilization.
For iron solubilization studies by measurement of soluble radioiron equilibrated with extrinsic stable iron, prepared aliquots of 3KR, 3KF, 1KR and 1KF in EBSS/MES were brought to iron concentration of 20 µmol/L by addition of ferric chloride stock solution. For comparison purposes, iron solubilization by ascorbic acid and nitrilotriacetic acid also was assessed: ferric chloride (20 µmol/L) was added to EBSS/MES at pH 6.0 with or without added ascorbic acid (1:20 iron/ascorbate molar ratio) or NTA (1:5 iron/NTA molar ratio). Iron solubility reference solution contained ferric chloride (20 µmol/L in 10 mmol of HCl/L, pH 2.0). 59FeCl3 (277.5 Bq) was added to aliquots (1.5 mL) of all solutions, which were vortex-mixed and allowed to sit 30 min at ambient temperature, with subsequent centrifugation for 30 min at 15,000 x g at ambient temperature (20
). The supernatant solutions were used for measurement of soluble radioiron.
For characterization of iron solubilization by using SEC/ICPMS before and after proteolysis, instrumentation consisted of the following components in series connected by PEEK and polyvinylchloride tubing as appropriate for high and low pressures, respectively: a) Model 501 Solvent Delivery System (Waters Chromatography Division, Millipore, Milford, MA); b) Model U6K Universal Liquid Chromatograph Injector (Waters Chromatography Division, Millipore); c) Prepacked Superdex Peptide Column, Model PE 7.5/300 (Amersham Biosciences, Piscataway, NJ); Model 484 Tunable Absorbance Detector (Waters Chromatography Division, Millipore); d) Model AT+ Ultrasonic Nebulizer (Cetac Technologies, Omaha, NE); and e) PlasmaQuad-3 inductively coupled plasma mass spectrometer (Thermo Elemental, Franklin, MA). Manganese ICPMS standard (SPEX Industries, Metuchen, NJ) was diluted with 0.1% vol/vol aqueous HCl to a concentration of 18.2 nmol/L and was pumped continuously at 0.4 mL/min with a peristaltic pump (Perimax 12; Spetec GMBH, Erding, Germany) through a PEEK tee fitting into the effluent from the absorbance detector. Data from the absorbance detector was processed by a Model HP3396 Series II Integrator (Hewlett Packard, Avondale, PA).
Elution solvent (MES, 0.1–0.3 mmol/L, pH 6.5–7.0) for the SEC/ICPMS system was delivered by the solvent delivery system at constant flow rates of 0.5 and 0.7 mL/min, so that column pressure was always <1,400 kPa. All injections were 50-µL aliquots of dilute aqueous solutions of samples and standards. Standard proteins and peptides with defined molecular weight and/or trace metal content were human apotransferrin (T-2252; Sigma Chemical), human milk -lactalbumin (L-7269; Sigma Chemical), bovine lung aprotinin (A-4529; Sigma Chemical, oxidized insulin chain B (I-6383; Sigma Chemical), crystalline cobalamin (V-2876; Sigma Chemical) and desferrioxamine mesylate (D-9533; Sigma Chemical). A molecular weight calibration curve was constructed by plotting the logarithm of the molecular weights against the partition coefficients Kav (21
) for aliquots of aprotinin, insulin chain B, cobalamin and desferrioxamine mesylate injected separately. The molecular weight for the major absorbance peak of the 1KR fraction was estimated graphically from the molecular weight calibration curve by identification of the logarithm of molecular weight that corresponded to the measured Kav for that peak.
The ICPMS instrument was operated in PlasmaScreen mode at forward power of 700 W with argon flow rates of 13.0 L/min (coolant), 0.42 L/min (auxiliary) and 1.03 L/min (external nebulizer); pole bias of -5.0 V; and with the following lens settings in volts: L2, 0.5; L3, -2.8; L4, -82.0; collector, -2.0; and extraction, -51. PlasmaLab software (version 1.06.00) was used for instrument control and for data display with acquisition parameters selected for transient time-resolved analysis of 16- to 25-min duration and with the following nominal mass-to-charge ratios (m/z) and dwell times (in milliseconds), respectively: 50, 249; 54, 249; 55, 10; 57, 249; 59 and 249. These nominal m/z correspond to monoisotopic Mn (m/z = 55) and Co (m/z = 59) and to the most interference-free isotopes of Cr (m/z = 50) and Fe (m/z = 54, m/z = 57).
In vitro digestion.
Proteolytic digestions were conducted as previously described (17
,22
) but with the following modifications for simulation of gastrointestinal conditions in infants, for microanalysis in polypropylene microcentrifuge tubes and without the addition of radioisotopic iron. Milligram quantities of lyophilized human -lactalbumin control and lyophilized 1KR fraction were dissolved in 1.1 mL of aqueous HCl, pH 2.70. Half of each solution was subjected to peptic incubation (60 U of pepsin in 1 µL of 0.1 mol of HCl/L were added) followed by slow addition of 2–4 µL of sodium bicarbonate (1 mol/L) to bring pH to 6.6. The other half of each solution was brought to pH 6.6 directly without prior peptic digestion. After removal of 50 µL from each tube for immediate analysis by SEC/ICPMS, incubation with pancreatin/bile extract mixture occurred (13 µg of pancreatin plus 80 µg of bile extract in 5 µL of 0.1 mol of NaHCO3/L was added to each tube, bringing pH to 7.0–7.4). Subsequent to centrifugation at 4°C as described (17
) but in a microcentrifuge (Eppendorf Model 5415C; Brinkmann Instruments, Westbury, NY), an aliquot of each supernatant fluid was withdrawn for immediate analysis by SEC/ICPMS. Then, aqueous FeCl3 (50 µL of 1 mmol of Fe/L, pH 2.7) was added to the remaining 0.45 mL of each supernatant of 1KR digests (pH of mixture, 6.9–7.0). Contents of tubes were vortex-mixed, incubated and subjected to centrifugation as described for measurement of soluble radioiron (20
), except that radioiron was not added. Finally, aliquots of these supernatants were analyzed for soluble iron and protein by SEC/ICPMS.
Cell culture.
Caco-2 cells were cultured as in previous reports (17
,22
) except that cells were obtained at passage 17 and used in experiments at passages 24–30, and fetal bovine serum was present at 10% vol/vol. The iron concentration of the prepared medium was 10 µmol/L; therefore, all cell monolayers consisted of iron-replete cells. Existence of confluent monolayers with tight intercellular junctions was verified by phenol red testing (17
).
Transmonolayer flux of iron.
Relationships between components of the Caco-2/transwell system are illustrated in Figure 2
. Procedures and conditions were as described previously (17
) except that apical buffer was EBSS/MES, pH 6.0, and basolateral buffer was EBSS/HEPES, pH 7.4. In all experiments, 59FeCl3 (833 Bq/4.5 mL preincubation volume) was added to apical buffer (blank), ferrous ascorbate control solution and prepared milk fractions 10 min before incubation with cell monolayers. In some experiments, extrinsic stable iron as ferric chloride stock solution was added (final extrinsic iron concentration, 15 µmol/L) concomitantly with the radioiron. Radioactivity in cell lysates was considered to represent radioiron cellular content plus binding, and radioactivity in basolateral buffer was considered to represent net basolateral transfer.
|
In a manner identical to the control incubations of the studies of transmonolayer flux, Caco-2 cell monolayers were exposed apically to ferrous/ascorbate control solution and 59Fe for 1 h at 37°C. Then, the apical and basolateral solutions were removed, cells were rinsed with 1 mL of EBSS and the solutions were saved for measurement of radioactivity. Monolayers were then incubated for 1 h with apical 3KF or 3KR but without added 59Fe. Control for these incubations was EBSS/MES without added 59Fe. Fresh (i.e., nonradioactive), EBSS/HEPES, pH 7.4, was placed in the basal chambers. After incubation, apical and basolateral solutions and cell monolayers were counted for radioactivity.
Statistical Analysis.
Analysis of the data were performed by using one-way ANOVA. Tukey’s multiple comparison test (Prism; GraphPad, San Diego, CA) was used to compare group means. Treatment comparisons were considered statistically significant when probabilities were <0.05.
|
RESULTS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
The 10KF fraction of mature human whey possessed 39% of the iron (iron concentration: 2.68 ± 0.18 µmol/L, mean ± SEM, n = 3) native to human milk (iron concentration: 5.75 ± 0.14 µmol/L) and the highest iron/protein ratio (67 µg/g) of the fractions analyzed (OFGM, 10KR, 10KF, casein). The OFGM fraction of the milk was second highest in intrinsic iron content (31% of milk iron) and highest in iron concentration (28.6 ± 1.2 µmol/L), followed by the 10KR fraction of whey at 26% of milk iron (17.9 ± 1.8 µmol/L). The casein fraction contained 2% of the protein and 1.4% of the iron (9.8 ± 1.8 µmol/L).
Radioiron solubility.
The solubility of radioiron in the presence of whey fractions in EBSS/MES and with stable iron concentration brought to 20 µmol/L by the addition of ferric chloride is illustrated in Figure 3
. Results are expressed as percentage of the radioiron solubility in reference ferric chloride (20 µmol/L) in 10 mmol of HCl/L, pH 2.0. Radioiron solubility in ferric chloride in EBSS, pH 6.0, without whey fractions was minimal (22.0 ± 0.2%). Solubility with 1KF added to the EBSS was only slightly, but significantly, greater (30.0 ± 0.3%). Radioiron solubilities in the presence of ascorbate or 1KR in EBSS were 80% of the reference value in dilute HCl and solubilities in the presence of NTA, 3KF or 3KR were 90–94% of reference.
|
and 5
. Integration of ultraviolet absorbance versus time at wavelength of 280 nm revealed that lyophilized, resuspended 1KR contained two broad peaks of 77 and 14 area%, respectively (chromatogram C, Fig. 4
). When fit to the molecular weight calibration curve, the partition coefficient for the larger peak corresponded to an apparent molecular weight of 4.2 kDa. After in vitro enzymatic digestion of 1KR, the total chromatogram area (chromatogram D, Fig. 4
) decreased to 50% of the chromatogram area of native 1KR (chromatogram C, Fig. 4
), and the larger peak decreased to 52 area%, whereas a second peak at longer elution time comprised 39 area%. Size exclusion chromatography of native human -lactalbumin (chromatogram A, Fig. 4
) revealed a single, sharp peak of 100 area% with an elution time shorter than those of the broad peaks from 1KR. In vitro enzymatic digestion of -lactalbumin decreased total chromatogram area (chromatogram B, Fig. 4
) to 27% of the area of chromatogram A, and the products of proteolysis were distributed over a wide range of elution times.
|
|
) at an elution time that corresponded to the maximum for ultraviolet absorbance of the larger 1KR peak and a similarly shaped intensity versus time curve at m/z = 57 (data not shown). The ratios of the intensities for m/z = 57 versus m/z = 54 at corresponding time points are similar to the natural abundance ratio for 57Fe/54Fe. The 1KR in solution also exhibits a small peak of intensity at m/z = 59 at an elution time similar to that observed for the cobalamin standard, as well as two very small peaks at m/z = 50, at different elution times from the aforementioned peaks, that may be ascribable to the presence of chromium complexes of low molecular weight (data not shown). The addition of FeCl3 to 1KR solution results in very intense, similarly-shaped peaks at m/z = 54 (chromatogram D, Fig. 5
) and m/z = 57 (data not shown) with ratios of intensities similar to the natural abundance ratio for the isotopes of iron observed at these nominal m/z values, and the plots for m/z = 50 and m/z = 59 are unaffected by addition of FeCl3 (data not shown).
Effects of digestive treatment on the iron content and iron-binding capacity of 1KR are illustrated (chromatograms A, B and C, Fig. 5
). After change in pH and incubation with pancreatin/bile extract, peak intensities at m/z = 54 (chromatogram B, Fig. 5
) and at m/z = 57 are only slightly decreased and peak shapes are skewed to longer elution times relative to pretreatment chromatograms (e.g., chromatogram E, Fig. 5
). The addition of incubation with pepsin to the treatment procedure before incubation with pancreatin/bile extract caused a further decrease in peak intensities, but peak heights and areas remained substantial (chromatogram C, Fig. 5
). Chromatogram A (Fig. 5)
illustrates that the addition of ferric chloride after change in pH and treatment with pepsin, pancreatin and bile extract resulted in almost as much bound iron with the same peak shape and elution time as does addition of FeCl3 to untreated 1KR (chromatogram D, Fig. 5
). ICPMS chromatograms of human -lactalbumin at m/z = 50, 54 (left inset, Fig. 5
), 57 and 59 establish that lactalbumin contains very little iron and confirm that elution time for lactalbumin-associated transition metals differs from that of 1KR-associated transition metals.
Radioiron flux through Caco-2 cell monolayers.
Integrity of the Caco-2 cell monolayers during incubation with milk fractions was studied by inclusion of phenol red (100 µmol/L) in apical chambers of some of the replicates of all the treatments shown in Figures 6
and 7
. Less than 2% of the dye appeared in the basolateral chambers after 1-h incubation at 37°C. Results for radioiron with and without phenol red were consistent with each other. Also, the integrity of tight junctions after incubation was confirmed for selected monolayers by electron microscopy and by measurements of transepithelial electrical resistance.
|
|
. The percentage radioiron content plus binding (means ± SEM) of Caco-2 cell monolayers postincubation was highest in the presence of ferrous ascorbate (FeASC). Of all the milk fractions, OFGM and 3KR elicited the highest radioiron content plus binding. Radioiron content plus binding in the presence of 10KF was slightly greater than that of the blank and 3KF. Means for radioiron content plus binding in the presence of milk, casein, 10KR, 1KR and 1KF were lower than that of the blank. In this set of experiments, the extrinsic apical iron contributed by the 59Fe was 5.73 nmol/L. The apical intrinsic iron concentrations of the various treatments were in the range of 1–10 µmol/L, which is commonly used in Caco-2 cell studies of iron flux (in µmol/L): milk, 1.2; OFGM, 1.1; 10KR, 1.1; 10KF, 2.4; 3KR, 1.4; 3KF, 4.9; 1KR, 5.4; 1KF, 2.5; casein, 0.84; blank, 0; FeASC, 10.0.
Basolateral radioiron transfer (Fig. 6)
in the presence of 10KF, 3KF and 1KR was about fivefold, fourfold and fourfold, respectively, that of the FeASC control when reported as percentage of total radioactivity. If equilibration of the radiolabel with intrinsic iron in each fraction is assumed, these mean values (in pmol of iron/well) are for 10KF, 270; 3KF, 448; 1KR, 446; and FeASC, 219. Therefore, iron transfer is apparently greater for these fractions than for ascorbate, even though the apical iron concentrations in the presence of these fractions are substantially lower than the apical iron concentration in the presence of ascorbate in this experiment. Values for radioiron transfer in the presence of milk and 3KR were not significantly different from that of FeASC, and OFGM and 1KF elicited slightly greater transfer. Transfer in the presence of casein and 10KR was significantly less than that of all foregoing treatments except 3KR. Casein, 10KR and 3KR elicited transfer not significantly different from that of the blank.
Radioiron flux in monolayers preincubated with stable and radioiron.
Preincubation of Caco-2 cells with ascorbate, radioiron and iron concentration at 10 µmol/L in apical medium (Fig. 7)
resulted in radioiron content plus binding of 12–14% of total radioactivity as in Figure 6
. Basolateral radioiron transfer after preincubation with ferrous ascorbate and 59Fe was <1% of total radioactivity, as in Figure 6
. After radioiron-free 3KF and 3KR were incubated with these Caco-2 cell monolayers that already bound/contained radioiron, values for mean monolayer radioiron content plus binding in the presence of 3KF and 3KR were 5.05 ± 0.05% and 6.08 ± 0.04%, respectively, of total radioactivity compared with 10.8 ± 0.1% for EBSS/MES (Fig. 7)
. The means for basolateral radioiron transfer after incubation with 3KF (7.49 ± 0.05%) and 3KR (5.67 ± 0.13%) were significantly greater than that after incubation with EBSS/MES alone (3.22 ± 0.10%).
|
DISCUSSION |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
. Ultrafiltration membranes do not have extremely sharp molecular weight cutoffs; they will pass appreciable amounts of molecules with weights up to two times the nominal cutoff value of the membrane but do not normally retain substantial amounts of molecules with weights below the nominal cutoff value. Therefore, the results illustrated in Figure 3
support solubilization of ferric iron by a human whey component of 1–6 kDa. The results with 1KF in Figures 3
and 6
are consistent and support those of a previous report that free citrate can decrease iron uptake by Caco-2 monolayers when the concentration of citrate is greater than that of ascorbate or cysteine, as is the situation in human milk (23
).
Ferric iron is almost as well solubilized by 1KR after as before pH change and in vitro enzymatic digestion (Fig. 5)
. Size exclusion chromatography of 1KR before and after in vitro enzymatic digestion (Figs. 4
and 5
) shows a broad ultraviolet absorbance peak with an elution time that corresponds to a molecular weight of 4.2 kDa and that co-chromatographs with iron content and iron-binding activity but not with human -lactalbumin. The observations that the ultraviolet absorbance of this broad peak is only moderately decreased (while that of the -lactalbumin positive control is severely reduced) on enzymatic treatment and that the iron content and iron-binding capacity are not proportionately diminished suggest that an iron-binding component is even more resistant to digestion than are other components of 1KR. Taken together, Figures 3
4
5
6
strongly suggest that a digestion-resistant whey component of 4.2 kDa enhances solubilization of nonheme, ferric iron at intestinal pH in support of increased iron absorption.
The basolateral radioiron transfer by Caco-2 cell monolayers in the presence of unfractionated breast milk is similar to that for ferrous ascorbate (Fig. 6)
, which is the customary positive control for extrinsic iron bioavailability experiments in Caco-2 cells. Relative to ferrous ascorbate, some milk fractions adversely affect, and others promote, radioiron flux through Caco-2 monolayers (Fig. 6
7)
. For examples, the casein and 10KR fractions resulted in basolateral transfer not significantly greater than blank and significantly less than all treatments except 3KR (Fig. 6)
. Our interpretation is that these unhydrolyzed casein- and lactoferrin-containing fractions of human milk can interact with extrinsic iron to minimize its uptake from apical medium. Whether this holds true for enzymatic digests of these fractions is not yet known.
In contrast to the casein and 10KR fractions, the 10KF and the low molecular weight whey fractions derived therefrom (except for 3KR) elicit severalfold more radioiron transfer than do ferrous ascorbate and unfractionated breast milk (Fig. 6)
. These observations, combined with the property of ultrafiltration membranes to pass molecules of mass up to twice the nominal molecular weight cutoff value, suggest that a whey component of <6 kDa facilitates basolateral transfer of iron.
The interpretation that a whey component of <6 kDa facilitates basolateral transfer of iron is reinforced by the results of the iron-loading experiments (Fig. 7)
. More than half of the radioiron in monolayers preincubated with ferrous ascorbate is transferred to basolateral medium when the monolayers subsequently are incubated with 3KF in the apical medium, but less than half of the radioiron is transferred during incubation with an equal amount of protein from 3KR. Kinetic studies in isolated duodenal segments and in Caco-2 cell monolayers have shown that basolateral transfer of iron is increased by subsequently absorbed chelators of iron such as EDTA and citrate anions (10
,24
). We suggest that iron-binding peptides from breast milk can be taken up by enterocytes and can mobilize iron from the intracellular labile pool for basolateral transfer.
The possibility must be considered that milks from other species possess low molecular weight components that enhance basolateral iron transfer. The iron distribution in bovine whey is similar to that in human milk whey in that about one half is bound to whey proteins and the other half is associated with low molecular weight compounds (6
). However, there are more differences than similarities between protein composition of milks of these species (25
). In bovine whey, ß-lactoglobulin predominates over -lactalbumin. Human milk contains no ß-lactoglobulin, and -lactalbumin is the predominant whey protein. Nonetheless, bovine skim milk exhibits protein bands on urea-PAGE that are of lower molecular weight than -lactalbumin and are not observed in the whole casein fraction (26
).
We conclude that low molecular weight whey fractions of breast milk enhance flux of extrinsic ferric iron across Caco-2 cells. We speculate that one or more peptides from these fractions resistant to digestion may enhance absorption of nonmilk iron in the infant proximal small intestine. Increased understanding of this phenomenon is expected to lead to development of ferric compounds that are more effective as iron fortificants for populations at risk and are more stable than commonly used ferrous forms that can be readily oxidized.
|
FOOTNOTES |
|---|
2 This work was supported in part by the Office of the Vice Provost for Research and Advanced Studies and by the Center for Designing Foods to Improve Nutrition of Iowa State University.
4 Abbreviations used: EBSS, Earle’s balanced salt solution without bicarbonate; FeASC, ferrous ascorbate; ICPMS, inductively coupled plasma mass spectrometry; 1KF, 1-kDa filtrate; 3KF, 3-kDa filtrate; 10KF, 10-kDa filtrate; 1KR, 1-kDa retentate; 3KR, 3-kDa retentate; 10KR, 10-kDa retentate; MES, 2-(N-morpholino)ethanesulfonic acid; NTA, nitrilotriacetic acid; OFGM, outer-fat-globular membrane; PEEK, polyethyletherketone; SEC, size exclusion chromatography.
Manuscript received 18 June 2002. Initial review completed 9 July 2002. Revision accepted 29 October 2002.
|
LITERATURE CITED |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
1. Saarinen, U. M., Siimes, M. A. & Dallman, P. R. (1977) Iron absorption in infants: high bioavailability of breast milk iron as indicated by the extrinsic tag method of iron absorption and by the concentration of serum ferritin. J. Pediatr. 91:36-39.[Medline]
2. Fomon, S. J., Ziegler, E. E., Nelson, S. E., Serfass, R. E. & Frantz, J. A. (1995) Erythrocyte incorporation of iron by 56-day-old infants fed a 58Fe labeled supplement. Pediatr. Res. 38:373-378.[Medline]
3. Abrams, S. A., Wen, J. & Stuff, J. E. (1996) Absorption of calcium, zinc and iron from breast milk by five-to-seven-month-old infants. Pediatr. Res. 39:384-390.
4. McMillan, J. A., Oski, F. A., Lourie, G., Tomarelli, R. M. & Landaw, S. A. (1977) Iron absorption from human milk, simulated human milk and proprietary formulas. Pediatrics 60:896-900.
5. Lönnerdal, B. (1984) Iron and breast milk. Stekel, A. eds. Iron Nutrition in Infancy and Childhood 1984:95-117 Raven Press New York. .
6. Fransson, G. & Lönnerdal, B. (1983) Distribution of trace elements and minerals in human and cow’s milk. Pediatr. Res. 17:912-915.[Medline]
7. Davidsson, L., Kastenmayer, P., Yuen, M., Lönnerdal, B. & Hurrell, R. F. (1994) Influence of lactoferrin on iron absorption from human milk in infants. Pediatr. Res. 35:117-124.[Medline]
8. Alvarez-Hernandez, X., Nichols, G. M. & Glass, J. (1991) Caco-2 cell line: a system for studying intestinal iron transport across epithelial cell monolayers. Biochem. Biophys. Acta 1070:205-208.[Medline]
9. Alvarez-Hernandez, X., Smith, M. & Glass, J. (1998) The effect of apotransferrin on iron release from Caco-2 cells, an intestinal epithelial cell line. Blood 91:3974-3979.
10. Halleux, C. & Schneider, Y. (1994) Iron absorption by Caco2 cells cultivated in serum-free medium as in vitro model of the human intestinal epithelial barrier. J. Cell. Physiol. 158:17-28.[Medline]
11. 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 human intestinal cell line. J. Nutr. 124:580-587.
12. Han, O., Failla, M. L., Hill, A. D., Morris, E. R. & Smith, J. C., Jr (1995) Reduction of Fe(III) is required for uptake of nonheme iron by Caco-2 cells. J. Nutr. 125:1291-1299.
13. Han, O., Failla, M. L. & Smith, J. C., Jr (1997) Transferrin-iron and proinflammatory cytokines influence iron status and apical iron transport efficiency of Caco-2 intestinal cell line. Nutr. Biochem. 8:585-591.
14. Arrendondo, M., Orellana, A., Garate, M. A. & Núñez, M. T. (1997) Intracellular iron regulates iron absorption and IRP activity in intestinal epithelial (Caco-2) cells. Am. J. Physiol. 273(2 Pt. 1):G275-G280.
15. Núñez, M. T., Núñez-Millacura, C., Beltran, M., Tapia, V. & Alvarez-Hernandez, X. (1997) Apotransferrin and holotransferrin undergo different endocytic cycles in intestinal epithelial (Caco-2) cells. J. Biol. Chem. 27:19425-19428.
16. Han, O., Fleet, J. C. & Wood, R. J. (1999) Reciprocal regulation of HFE and Nramp2 gene expression by iron in human intestinal cells. J. Nutr. 129:98-104.
17. Au, A. P. & Reddy, M. B. (2000) Caco-2 cells can be used to assess human iron bioavailability from a semipurified meal. J. Nutr. 130:1329-1339.
18. Hamosh, M., Ellis, L. A., Pollock, D. R., Henderson, T. R. & Hamosh, P. (1996) Breastfeeding and the working mother: effect of time and temperature of short-term storage on proteolysis, lipolysis, and bacterial growth in milk. Pediatrics 97:492-498.
19. Lowry, O. H., Rosebrough, N. J., Farr, A. L. & Randall, R. J. (1951) Protein measurement with the Folin phenol reagent. J. Biol. Chem. 193:265-275.
20. Glahn, R. P., Wien, E. M., Van Campen, D. R. & Miller, D. D. (1996) Caco-2 cell iron uptake from meat and casein digests parallels in vivo studies: use of a novel in vitro method for rapid estimation of iron bioavailability. J. Nutr. 126:332-339.
21. Kuchler, R. J. (1972) Design of chromatographic procedures. Umbreit, W. W. Burris, R. H. Stauffer, J. F. eds. Manometric & Biochemical Techniques 5th ed. 1972:336 Burgess Publishing Minneapolis. .
22. Swain, J. H., Tabatabai, L. B. & Reddy, M. B. (2002) Histidine content of low-molecular-weight beef proteins influences nonheme iron bioavailability in Caco-2 cells. J. Nutr. 132:245-251.
23. Glahn, R. P., Lai, C., Hsu, J., Thompson, J. F., Guo, M. & Van Campen, D. R. (1998) Decreased citrate improves iron availability from infant formula: application of an in vitro digestion/Caco2 cell culture model. J. Nutr. 12:257-264.
24. Schümann, K., Elsenhans, B., Hunder, G., Strugala, G. & Forth, W. (1993) Impact of chelators on the bioavailability of iron after its uptake into the mucosa. Schlemmer, U. eds. Bioavailability ’93 Nutritional, Chemical, and Food Processing Implications of Nutrient Availability Proceedings 1993:33-37 Bundesforschungsanstalt für Ernährung Karlsruhe. .
25. Heine, W. E., Klein, P. D. & Reeds, P. J. (1991) The importance of -lactalbumin in infant nutrition. J. Nutr. 121:277-283.
26. Van Hekken, D. L. & Thompson, M. P. (1992) Application of PhastSystem to the resolution of bovine milk proteins on urea-polyacrylamide gel electrophoresis. J. Dairy Sci. 75:1204-1210.[Abstract]
This article has been cited by other articles:
|
|
 |
|
  P. Etcheverry, D. D. Miller, and R. P. Glahn A Low-Molecular-Weight Factor in Human Milk Whey Promotes Iron Uptake by Caco-2 Cells J. Nutr., January 1, 2004; 134(1): 93 - 98. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
8) with different letters are significantly different (P < 0.05). Some error bars are too small to be visible. Results are expressed as percentage of total radioactivity. OFGM, outer-fat-globular membrane.