Heme-Fe is an important source of dietary iron in humans. Caco-2 cells have been used extensively to study human iron absorption with an emphasis on factors affecting nonheme iron absorption. Therefore, we examined several factors known to affect heme iron absorption. Cells grown in bicameral chambers were incubated with high specific activity [59Fe]heme alone or with 1% globin, BSA, or fatty acid-free BSA (BSA-FA) to examine the effect of protein source on absorption. Heme iron absorption was enhanced by globin and inhibited by BSA and BSA-FA. Absorption of heme iron in cells pretreated for 7 days with serum-free medium containing 1, 25, 50, or 100 μM Fe was higher in the 1-μM-Fe pretreatment group than in all other groups (P < 0.05), showing an effect of iron status. Increased heme concentrations resulted in decreased percent absorbed but increased total heme iron absorption and increased transport rate across the basolateral membrane. Finally, cells treated with 10 μM CdCl2, which induces heme oxygenase, demonstrated higher absorption of [59Fe]heme than control cells (P < 0.05). Our results from Caco-2 cells are in agreement with human studies and make this a promising model for examining intestinal heme iron absorption.
- heme oxygenase
- cell culture
there are relatively few studies describing the mechanisms behind the process of intestinal heme iron absorption despite the importance of heme iron as a highly bioavailable source of dietary iron. Populations that consume meat as a significant component of their diet are normally iron replete. In fact, it has been determined that two-thirds of absorbed dietary iron in North Americans and Europeans is derived from heme iron, although it only comprises one-third of dietary iron (22, 38). Intestinal absorption of heme iron is higher than that of nonheme iron, suggesting that heme may be a preferred iron source in iron deficiency; it may also be a source of dietary iron to avoid when iron status is high, such as in hemochromatosis (20, 26, 35, 36).
Until recently, most isotopic studies on heme iron absorption were performed in vivo in animals or humans or in vitro in isolated intestine. Human studies have proven useful in elucidating the effects of dietary components on heme iron absorption. Hallberg et al. (21) demonstrated inhibition of heme iron absorption in the presence of calcium in humans. An enhancing effect of globin on heme iron absorption has also been described with more heme iron absorbed in the presence of globin than with heme alone (12). Such human studies, however, have their limitations. One limitation is the synthesis of high specific activity radiolabeled heme. The most common method is in vivo (intrinsic) labeling in which an anemic animal is injected with high specific activity59Fe or 55Fe. The animal is bled, and hemoglobin is isolated from red blood cells. This method, however, requires time to induce anemia and yields only low to modest levels of specific activity. Another method is in vitro synthesis of14C-labeled heme iron, which labels the porphyrin ring but does not allow determination of iron retention in the body. Human studies are also expensive and time consuming, limiting their use in exploring the effects of specific dietary components. Therefore, the development of a less expensive screening method has been suggested (18).
The Caco-2 cell line has become one of the most utilized cell lines in intestinal iron research. This human colon adenocarcinoma cell line is capable of enterocytic differentiation and displays many structural and functional properties of mature human intestinal epithelial cells (19). The cells form a highly polarized monolayer of cells exhibiting tight junctions and apical microvilli with brush border membrane and are known to express enzymes specific to the enterocyte (10, 15). Caco-2 cells absorb and transport inorganic iron and, similar to results from human studies, ferrous iron is absorbed more efficiently than ferric iron (1). Inorganic iron absorption has also been shown to respond to dietary components in a similar manner to that observed in humans (3). The basolateral release of iron to apotransferrin seen in human intestinal epithelial cells also occurs in Caco-2 cells (2, 27, 28). The cells are sensitive to iron status; e.g., increasing the level of iron in the medium decreases iron uptake and transport (1,17, 34). Most recently, regulation of gene expression of two newly identified proteins involved in iron metabolism was examined in the Caco-2 cell model. It was shown that HFE and Nramp2 gene expression was altered by iron status of the cell, further supporting the use of Caco-2 cells as a model for iron absorption (23,33).
Although most studies on iron absorption by Caco-2 cells have focused on inorganic iron, it is known that, like human enterocytes, Caco-2 cells contain heme oxygenase, which is involved in the removal of iron from heme in the intestine, and that Caco-2 heme oxygenase expression and activity can be induced both by the presence of heme or heavy metals, such as cadmium (7). An increase in ferritin levels has also been seen in Caco-2 cells in the presence of heme, suggesting increased iron stores within the cell (9). In addition, recent work by Worthington et al. (40) may suggest the presence of a putative heme receptor in Caco-2 cells, but the actual transport mechanism has not been elucidated. In contrast, Liem et al. (25) suggest the process is independent of a receptor. Instead, heme intercalates into the hydrophobic environment of the lipid bilayer.
The following study was undertaken to examine the appropriateness of using Caco-2 cells in culture for studies on heme iron absorption and transport. Radiolabeled heme was synthesized in vitro using a method that allows rapid production of [59Fe]heme without the use of an animal and with consistent high specific activity.
MATERIALS AND METHODS
Radiolabeled iron was purchased from Amersham (Arlington Heights, IL). All solvents used in the synthesis of [59Fe]heme were obtained from Fisher Scientific. All other supplies were purchased from Sigma (St Louis, MO) unless stated otherwise.
Caco-2 cells (HTB37) from American Type Culture Collection (Manassas, VA) were maintained in MEM (GIBCO, Gaithersburg, MD) supplemented with 10% FBS (Gemini Bioproducts, Calabasas, CA) and antibiotics (100 U/ml penicillin and 100 U/ml streptomycin). Cells were cultured at 37°C in an incubator with a 5% CO2-95% air atmosphere maintained at constant humidity. After reaching 80% confluency, the cells were reseeded onto Transwell bicameral chambers with a 0.4-μm pore size membrane consisting of 12 wells per plate (Costar, Cambridge, MA) and grown to confluency. Cells were used between the 45th and 50th passages.
All experiments were carried out with confluent cells cultivated 18–21 days on bicameral chambers. Before each experiment, confluency of the monolayer was assayed. Formation of an intact monolayer was monitored by measuring the transepithelial electrical resistance with a Millicell electrical resistance system (Millipore, Bedford, MA). Iron uptake studies were performed after transepithelial electrical resistance measurements >250 Ω/cm2 were obtained, indicating formation of an intact monolayer (22,33, 34).
Preparation of [59Fe]heme iron.
[59Fe]heme was synthesized in vitro using a method developed by Dr. Jerry Bommer (Frontier Scientific, Logan, UT, personal communication) with minor modifications to accommodate for the radioactive iron being in the ferric state. Briefly, 20 ml of glacial acetic acid was heated in a water bath to 50°C under a continuous stream of nitrogen. Then, 16.25 mg of protoporphyrin IX dimethyl ester (Frontier Scientific) dissolved in a mixture of 2.75 ml of chloroform and 1 ml of pyridine was slowly added to the stirred glacial acetic acid. [59Fe]ferric chloride (100 μCi) in 0.1 M HCl (Amersham) dissolved in 100 μl of 17 mM ascorbic acid was added to the acetic acid/protoporphyrin solution and allowed to react for 1.5 h at 50°C. Then, 1.5 mg of cold ferrous sulfate dissolved in a minimum amount of methanol (500 μl) and acetic acid (50 μl) was added to the solution and incubated for 2 h. This solution was allowed to cool to room temperature, chloroform (50 ml) was added, and the mixture was washed in a separatory funnel with 3 × 15 ml of water to remove acetic acid and pyridine. The chloroform fraction was then transferred to a round-bottom flask and evaporated to ∼2.5 ml under reduced pressure and diluted with 12.5 ml of ethyl acetate. The solution, now containing hemin dimethyl ester, was washed in a separatory funnel with 4 N HCl to remove unincorporated protoporphyrin dimethyl ester until the acid washes were colorless. The organic fraction was then washed with water and evaporated to dryness under reduced pressure. Solid residue was dispersed in 2.5 ml of tetrahydrofuran and stirred with 0.5 ml of 3 N NaOH under nitrogen for 12–24 h (until all hemin had salted out). The NaOH solution was diluted with 1 ml of water, and the product was precipitated by the addition of 10 N HCl dropwise while stirring. Precipitated [59Fe]hemin was collected by centrifugation, washed twice with water, and dried overnight by vacuum dessication. Purity was determined by comparing the absorbance of the synthesized hemin at 320–420 nm with commercial hemin and the initial starting product, protoporphyrin IX dimethyl ester. Concentration was determined by comparison of absorbance of [59Fe]heme in the Soret wavelength region (390 nm) with known standards. Standards were prepared by dissolving 1 mg of hemin in 50 μl of 1 M NaOH and further dilution in 10 mM potassium phosphate buffer (pH 7.45). Standards were always prepared fresh and kept from exposure to light. The [59Fe]heme specific activity was 6.7 Ci/mol heme.
Effects of protein on heme iron uptake.
In one series of experiments, the effects of human globin, BSA, and fatty acid-free BSA (Sigma) on heme iron absorption by Caco-2 cells was determined. Heme solutions were made immediately before introduction to the cells. In these studies, 50 μM of [59Fe]heme alone or combined with solutions of either 1% globin, albumin, or fatty acid-free albumin was administered to the apical surface of the Caco-2 monolayer. The [59Fe]heme solution was prepared by solubilization of 0.75 μCi [59Fe]heme and carrier heme in 10 μl of 0.1 N NaOH. Ninety microliters of 10 mM potassium phosphate buffer alone or buffer containing globin or albumin were added. The entire solution was brought up to 500 μl with serum-free MEM to obtain a final concentration of 50 μM heme and 1% globin, albumin, and fatty acid-free albumin. MEM with FBS was removed from the apical and basal chambers. Serum-free MEM containing 20 μM apotransferrin was added to the basal chamber, and 1 ml of the experimental [59Fe]heme solutions was added to the apical chamber of the Transwell.
Cultures were incubated at 37°C. At 6, 12, 24, and 48 h, all basal medium was removed for counting at each time point and replaced with fresh 25 μM apotransferrin in serum-free MEM. At 48 h, the medium covering the cells was removed and the cells were washed three times with PBS and the wash solution was collected for counting. Counts in the apical and wash solutions represented unabsorbed [59Fe]heme. After the monolayer was washed, the Transwell membrane on which the cell monolayer was attached was removed from the Transwell and placed in 0.1 N NaOH for 24 h to detach the cells from the membrane. This solution was counted by liquid scintillation (Wallac 1410 liquid-scintillation counter) for determination of59Fe within the cell. Counts obtained from the cell fraction along with counts found in the basal medium represented heme iron absorbed by the monolayer. Counts obtained from the basal medium alone represented 59Fe transport out of the cell and across the basal membrane of the monolayer.
Effect of iron status on heme uptake.
Another set of experiments was performed to examine the effect of cell iron status on heme iron absorption. Cells were incubated with varying amounts of iron (1, 25, 50, and 100 μM Fe) in serum-free MEM for 7 days before the addition of labeled [59Fe]heme. Iron was added to the medium as ferric nitrilotriacetic acid (NTA) at a 1:4 molar ratio. Then, 50 μM heme and 1% globin in serum-free MEM were added to the upper chamber of the Transwell and 20 μM apotransferrin in serum-free MEM was added to the lower chamber. Collection of data was the same as in the previous experiment.
Effect of heme concentration on heme uptake.
The effect of varying amounts of heme on heme iron absorption and transport by the cell was also explored. Varying concentrations of heme (1, 6, 25, and 75 μM) and 1% globin in serum-free MEM were introduced to the upper chamber of the Transwell and heme iron absorption was monitored over a 48-h period.
Effect of heme oxygenase induction on iron uptake and transport.
The final experiment was performed to examine the effect of cadmium on iron transport across the monolayer. Cells were pretreated with 10 μM cadmium (as CdCl2) in serum-free MEM for 10 h before the addition of heme. After pretreatment, 50 μM [59Fe]heme and 1% globin in serum-free MEM containing 10 μM cadmium were added, and transport of iron was measured as described above.
In all experiments, variables were tested in triplicate wells, and the means ± SE were determined. Each experiment was performed in a separate 12-well plate to avoid absorption variability between plates. The distribution of radioactivity between the apical and basolateral medium, cell wash, and cell homogenate was determined by liquid- scintillation counting. Values for accumulation of [59Fe]heme were converted to the percentage of the initial concentration of heme introduced to the cell that was found in the lower chamber solution and in the cell monolayer after 48 h. Knowing the molar concentration administered to the apical side and the determination of microcuries in the initial dose, allowed for the calculation of picomole per hour by counting of the basal medium at the designated time point.
Statview version 5.0.1 and KaleidaGraph version 3.0 (Synergy Software, Reading, PA) were used to analyze and graph the cell absorption and transport data. Outcome variables were analyzed by two-way ANOVA with main effects of time and protein source using Statview ANOVA with post hoc comparisons according to Fisher's protected least significant difference test to compare effects of treatment groups and time on heme iron absorption at the 0.05 level of significance. A t-test was used for the analysis of cadmium treatment on absorption also at the 0.05 level of significance.
Effect of globin and albumin on heme iron absorption.
In the first experiment, [59Fe]heme was added to the apical surface of the monolayer as heme alone or in the presence of 1% globin, BSA, or fatty acid-free albumin (fatty acid-free BSA). Fig. 1 A shows the transport of 59Fe over time across the basolateral side, whereas Fig. 1 B demonstrates the accumulation of59Fe at the basolateral side over a 48-h time period. The transport rate for 59Fe from heme + globin was greater than from heme alone, with BSA, or with fatty acid-free BSA at 24 and 48 h with rates of 2.2 ± 0.5 and 1.7 ± 0.2 vs. 1.4 ± 0.1 and 1.1 ± 0.02 pmol/h for heme alone (P < 0.5). The rate of accumulation of59Fe in the basolateral chamber over a 48-h time period was significantly higher in the presence of globin at 12, 24, and 48 h (P < 0.05) than in all other treatment groups with accumulation rates of 19.7 ± 5.8, 46. 2 ± 10.1, and 88.7 ± 13.6 pmol for the globin treatment group vs. 8.9 ± 2.9, 25.3 ± 3.8, and 50.8 ± 3.8 pmol for heme alone. Accumulation rates in the presence of BSA and fatty acid-free BSA were significantly lower at 12, 24, and 48 h than from heme alone (P < 0.05) with 2.8 ± 0.4, 7.2 ± 0.4, and 21.5 ± 1.2 pmol for BSA and 3.0 ± 1.4, 6.0 ± 1.5, and 17.0 ± 1.1 pmol for fatty acid-free BSA.
Effect of iron status on heme iron absorption.
Confluent monolayers were pretreated for 7 days with serum-free medium containing 1, 25, 50, or 100 μM iron (ferric NTA) to determine the effect of cellular iron status on heme iron absorption. After 7 days, [59Fe]heme in the presence of 1% globin was added to the apical surface of the monolayer. Percentages of 59Fe from the original dose contained within cells were 5.6 ± 1.4, 3.3 ± 0.1, 2.9 ± 0.2, and 2.6 ± 1.3% for cells pretreated with 1, 25, 50, and 100 μM iron, respectively (Fig.2). Cells pretreated with 1 μM iron showed higher concentrations of 59Fe from heme than cells pretreated with 25, 50, and 100 μM iron. Percentage of the original dose absorbed by cells and transferred to basal medium was higher in the 1 μM pretreatment group than in all other groups.
Effect of heme concentration on heme iron absorption.
Confluent monolayers were incubated with [59Fe]heme at varying concentrations of heme (1, 6, 25, and 75 μM) to determine the effect of dose on heme iron absorption. Transport increased rapidly in all groups with maximal transport of iron across the basolateral membrane occurring at 18 h. Figure3, A and B shows the transport rate in picomole per hour and percent per hour of the original dose in the 25- and 75-μM treatment groups. The 75-μM treatment group had the highest rate of iron transport across the basolateral membrane at all time points compared with the 25-μM group (P < 0.05). The highest transport from 75 μM heme was at 18 h with a rate of 9.7 ± 0.8 pmol/h compared with 4.4 ± 0.4 pmol/h from 25 μM heme. (Fig. 3 A). Although the molar amount of iron transported was higher in the presence of 75 μM heme, the percentage of the initial dose transported per hour across the basolateral membrane was lowest in the 75-μM group with a value of 0.03 ± 0.002%/h at 18 h vs. 0.04 ± 0.003%/h in the presence of 25 μM heme (P = 0.01) (Fig. 3 B).
Similar results were found when examining accumulation of59Fe in the lower chamber medium as picomole and as percentage of the original dose (data not shown). Although the 75-μM group shows a higher molar accumulation of 59Fe at all time points, the percentage of original dose actually transported across the basolateral membrane was lower than for the 25 μM treatment group (P < 0.05).
Percentages of the dose accumulated within cells after 48 h were 2.2 ± 1.1, 1.8 ± 0.5, 1.3 ± 0.1, and 0.6 ± 0.3% for 1, 6, 25, and 75 μM, respectively (Fig.4). Percentages of the original dose taken up by cells and transported into medium by cells incubated with 1, 6, 25, or 75 μM heme were 3.0 ± 1.4, 2.9 ± 0.6, 2.6 ± 0.1, and 1.6 ± 0.4, respectively. Percentage of original dose within cells and percent absorption in the 75 μM treatment group were found to be significantly lower than for the 6 and 25 μM groups.
Heme iron absorption as a function of heme oxygenase induction.
Confluent monolayers were pretreated for 10 h before the addition of heme with 10 μM cadmium in serum-free MEM. After pretreatment, [59Fe]heme and 1% globin in serum-free MEM containing 10 μM cadmium chloride were added. Percentage of the original dose of59Fe located within the cell and percentage of the total amount absorbed (cell concentration + basolateral accumulation) after 48 h were measured. As can be seen in Fig.5, a significantly higher amount of [59Fe]heme was found in cells treated with cadmium (5.7 ± 1.4%) than in the control group (3.5 ± 0.4%;P = 0.05). Absorption was also higher in the cadmium treated group with 8.6 ± 1.0 vs. 6.0 ± 0.6% of [59Fe]heme absorbed in the control cells (P = 0.02).
Effect of globin and albumin on heme iron absorption.
Heme iron in the diet is found predominantly in the form of hemoglobin and myoglobin. Conrad and colleagues (12-14) demonstrated increased absorption of heme iron given as hemoglobin or as globin degradation products compared with heme alone in both humans and animal models. The increased absorption was attributed to globin preventing heme aggregation through coordination bonding of the iron to a nitrogen in a histidine residue located in the globin chain (32). This hypothesis was tested by the replacement of globin with free histidine as well as niacin, which can also participate in coordination bonding with heme through nitrogenous ligands (8, 13, 24). Both free histidine and niacin were shown to decrease heme aggregation leading to a concomitant increase in iron absorption (12, 13). Our study in Caco-2 cells did find significant differences in transport and accumulation of [59Fe]heme when comparing cells treated with heme alone or in the presence of hemoglobin.
In addition, we found a pronounced decrease in transport, accumulation, and absorption of [59Fe]heme in the presence of albumin. Serum albumin serves as a temporary holding site for heme until it can be removed from circulation and degraded in the liver. Human serum albumin is capable of binding several moles of heme; however, there is only one primary binding site for heme with the other binding occurring through low-affinity hydrophobic binding. Binding of heme to albumin does not appear to involve the iron in heme, because it has been demonstrated that iron-free protoporphyrin can bind with the same affinity as heme iron (5). In addition, the iron in heme can still be reduced or bind to cyanide when heme is bound to albumin (5). Instead, there is evidence that binding occurs through the propionyl side chains of heme and involves a tryptophan residue in albumin.
Our results suggest that participation of iron and nitrogenous ligands in the binding of heme may be a more important component of intestinal heme iron absorption than just preventing heme aggregation. This is supported, in part, by findings in the bacteria Hemophilus influenzae in which binding of hemoglobin to the bacterial membrane could be inhibited by the presence of excess hemoglobin and globin but not by albumin (16). In addition, although heme in the body can be bound to albumin, albumin is not present in the gut and is not a source of heme iron in the diet (38). This suggests intestinal binding might more likely be adapted to the form of heme iron found in the predominant dietary sources, i.e., hemoglobin and myoglobin.
Effect of iron status on heme iron absorption.
Iron status of the individual has been shown to be an important factor in the regulation of iron absorption from the diet. The effect of iron deficiency on nonheme iron absorption is well documented with ≤12-fold increase in absorption seen in iron-deficient humans (6, 8, 11,22, 31, 38). Heme iron absorption has also been shown to be affected by iron status in humans with a three- to fourfold increase observed (8, 22, 31, 37, 38). Caco-2 cells have been shown to respond to iron status as well with increased uptake of nonheme iron seen when cells are grown in iron-deficient medium (1,34). Our results show a similar trend for an inverse relationship between iron status and heme iron absorption with the highest absorption of heme iron found in cells pretreated with serum containing 1 μM iron and the lowest absorption in cells pretreated with serum containing 100 μM iron. Removal of iron from heme is believed to be the rate-limiting step in heme metabolism. During iron deficiency, it is known that degradation of heme is increased by upregulation of heme oxygenase, thus increasing the amount of iron available for absorption (30). There is also some evidence as well that uptake of heme iron by the apical membrane of the epithelial cells is increased under condition of iron deficiency in the rat (31). Our results are in agreement with human and animal experiments showing increased absorption of heme during iron-deficient conditions.
Effect of heme concentration on heme iron absorption.
Animal and human studies have shown that the amount of heme iron absorbed is correlated to the amount ingested. The percentage of heme absorbed in both rats and guinea pigs decreased with increasing concentrations of heme, whereas the absolute amount of the dose absorbed increased (4, 23, 37, 39). Turnbull et al. (37) demonstrated similar results in humans; with larger doses of heme given, percent absorption decreased, whereas the total amount absorbed increased. Our results in Caco-2 cells confirm that heme absorption is dependent on dose. In addition, cells treated with increasing concentrations of heme iron also demonstrated an increase in the transport rate of 59Fe across the basolateral membrane and a decrease in the percentage of initial dose transported. The exact mechanism behind the increased transport rate is not clear; however, it is known that heme oxygenase activity is upregulated in the presence of heme (7). Therefore, increasing the amount of heme iron could lead to increased inorganic iron available for transfer across the basolateral membrane.
Heme iron absorption as a function of heme oxygenase induction.
The enzyme heme oxygenase is present in many tissues in the body, including the intestine, with the highest concentration observed in the duodenum (30). Heme oxygenase activity can be increased 10- to 100-fold by several conditions including hyperthermia, heme, and metal ions such as cobalt and cadmium (7, 29). We wanted to determine whether pretreatment of Caco-2 cells with cadmium would increase the amount of [59Fe]heme absorbed by the cell and the amount transported across the basolateral membrane. Our results show higher absorption and transport of 59Fe in cadmium treated cells, which should be consistent with increased activity of heme oxygenase. Heme oxygenase is known to be abundantly present in Caco-2 cells and its mRNA expression and activity have been shown to be increased in the presence of cadmium (7). This is the first study to report that induction of heme oxygenase activity by cadmium affects the amount of iron transported across the basolateral membrane of a Caco-2 cell monolayer.
In conclusion, our study shows that Caco-2 cells can be used as a model to study intestinal heme iron metabolism when high specific activity [59Fe]heme is used. Caco-2 cells showed increased transport and accumulation of radiolabeled heme iron in the presence of globin and an inhibition in the presence of albumin. Iron status of the cells was shown to affect heme iron absorption with an increase during iron deplete conditions similar to what is observed in whole animals. Changes in heme iron absorption, similar to those seen in whole animals, with more heme transported with increasing dose but a decrease in the percentage of the dose absorbed were also observed in our Caco-2 cell model. In addition, the presence of cadmium increased the transport of radiolabeled iron suggesting an increase in heme oxygenase activity. Our results suggest that Caco-2 cells in culture is a valid model for assessing heme iron transport from the intestinal lumen to the circulation. Ease of use, low cost, and confirmed use of these cells as a model for nonheme iron absorption make them a promising tool that may be used in combination with human studies to further examine factors and metabolic conditions affecting heme iron absorption.
We thank Rashida Rasheed for assistance with tissue culture.
Address for reprint requests and other correspondence: B. Lönnerdal, Dept. of Nutrition, Univ. of California, One Shields Ave., Davis, CA 95616 (E-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
July 11, 2002;10.1152/ajpgi.00443.2001
- Copyright © 2002 the American Physiological Society