GASTROENTEROLOGY 1990;96:576-565
Transferrin Receptor Distribution and Regulation in the Rat Small Intestine Effect of Iron Stores and Erythropoiesis GREGORY J. ANDERSON, LAWRIE W. POWELL, and JUNE W. HALLIDAY Department of Medicine, University of Queensland, Brisbane,Queensland, Australia
A combination of biochemical quantitation and immunohistochemistry has been used to examine in detail transferrin receptor distribution and expression in the rat small intestine and its relationship to iron absorption. Receptor numbers were quantitated by transferrin binding to preparations of basolateral or brush-border membranes. Receptors were demonstrated on the basolateral membranes of the gut cells, but not on the brush-border fraction. Apotransferrin demonstrated little binding to basolateral membranes at physiological pH. Dietary or parenteral iron loading of animals produced a significant decline in transferrin binding, whereas binding was increased in iron deficiency. These data were confirmed by immunohistochemical studies using a monoclonal antibody to the transferrin receptor. When iron absorption was increased threefold following acute hemolysis and without a decrease in body iron stores, there was no change in transferrin receptor number. These data indicate that intestinal transferrin receptors may be regulated by body iron stores but suggest that they are not directly involved in iron absorption.
he intestinal epithelial cell is one of the few cell types that transport iron vectorially. Iron must be taken into the cell across the brush-border membrane, transported through the cytoplasmic compartment, and released across the basal and/or lateral membranes into the extravascular fluid (1). Some of the details of brush-border uptake have been determined (2-6) and several intracellular iron-binding proteins have been identified (Y-10) but very little is known about the handling of iron at the serosal side of the enterocyte. A number of recent histochemical studies have shown transferrin receptors on the basolateral mem-
T
branes of intestinal epithelial cells (11-13). These have led to speculation on the role of the transferrin receptor in the gut mucosa and the role of transferrin in iron absorption (14). In addition, there is a proposal that transferrin is secreted into the intestinal lumen, binds iron, and is reabsorbed by a receptor-mediated process (15). It is not yet known whether enterocytes internalize transferrin by endocytosis, and there are few biochemical data on intestinal transferrin receptors, but because the receptor-mediated uptake of transferrin provides a major pathway of iron uptake by cells (16) the transferrin receptor may play an important role in the response of the intestinal cell to physiological stimuli. Most previous studies of intestinal transferrin receptors either have used semiquantitative immunohistochemical techniques or have been incidental observations made as parts of other studies. The purpose of the present report is to provide quantitative biochemical data on the distribution and numbers of intestinal transferrin receptors in relation to iron absorption. Using the rat as a model system we have been able to carry out a systematic study of the effects of body iron stores and erythropoiesis on intestinal receptor levels. These data suggest that intestinal transferrin receptor expression is related to iron stores, but that the receptor is not directly involved in iron absorption.
Materials and Methods Animals and Diets Adult male Sprague-Dawley rats were used for all experiments. Animals were divided into five groups:
Abbreviation used in this paper: BSA,bovine serum albumin. 0 1990 by the American Gashoenterological Association 0016-5065/90/$3.00
INTESTINAL TRANSFERRIN RECEPTORS 577
March 1990
1. Rat pellet-fed 2.
3.
4.
5.
animals received standard rodent pellet diet (determined iron content, 47 mg/kg wet wt). Iron-replete animals received the iron-deficient diet described by Valberg et al. (17) supplemented with ferrous ammonium sulfate (0.75 g/kg dry wt) (determined iron content, 159 mg/kg wet wt). Iron-deficient animals received the iron-deficient diet described by Valberg et al. (17) (determined iron content, 3 mg/kg wet wt). In addition, these animals were allowed access to deionized water only and were housed in all-plastic cages with glass water bottles. These conditions were essential for iron deficiency to be achieved and maintained. Carbonyl iron-supplemented animals received the irondeficient diet supplemented with 2.5% wt/wt (dry wt) carbonyl iron (GAF Corp., New York, N.Y.) (determined iron content, 14.9 g/kg wet wt). This diet produces hepatic parenchymal cell iron loading (18). Iron dextran-supplemented animals were maintained on the rat pellet diet but received intramuscular injections of 50 mg of iron dextran (Imferon; Fisons Pty Ltd., Sydney, Australia) in each of five successive weeks. This treatment produced iron-loaded animals in which iron was predominantly localized in the reticuloendothelial system.
The iron content of each diet was determined by atomic absorption spectrometry. Animals began receiving the irondeficient diet from the time of weaning, whereas all other treatments were started when the animals weighed 150 g. Animals in which reticulocytosis was induced [see below) received the standard rat pellet diet. All experiments described in this study were approved by the Animal Experimentation Ethics Committee of the university.
Transferrin
and Apotransferrin
Preparation
Transferrin was prepared from the pooled sera of normal rats by sequential ion exchange (Whatman DE-52 cellulose: Whatman Ltd., Maidstone, Kent, England) and size exclusion (Sephacryl S-200; Pharmacia [Australia] Pty. Ltd., Sydney, Australia) chromatography of an ammonium sulfate fraction precipitating between 40% and 70% saturation (7). Apotransferrin was prepared by dialyzing holotransferrin overnight against 50 mM sodium acetate (pH 5.0) containing 0.25 mM desferrioxamine (as mesylate [Desferal; Ciba Geigy Pty Ltd., Sydney, Australia]), followed by incubation with an equal volume of Chelex 100 (Bio-Rad Laboratories, Sydney, Australia) for 1 h at room temperature. The Chelex was removed by low-speed centrifugation.
Iron Absorption
Measurements
Iron absorption was measured by whole-body counting following administration of a tracer dose of 5gFe. For each 250-~1 dose, 0.5 &i of 59Fe ferric chloride (Amersham Australia Ltd., Sydney, Australia) was added to a solution of ferrous sulfate in 10 mM HCI to give a final iron concentration of 200 wg/ml. The dose was administered intragastritally to unanesthetized rats that had been fasted overnight. Whole-body radioactivity was determined 2-3 h and 7 days
after iron administration using a ARMAC large-volume gamma counter (Packard Instrument Company, Downers Grove, Ill.). Absorption was calculated as the percentage of iron remaining in the body 7 days after correction for radionuclide decay.
Induction
of Increased
Erythropoiesis
An increase in erythropoiesis followed reticulocytosis induced in 200-g rats by the subcutaneous injection of a single lo-mg dose of neutralized phenylhydrazine hydrochloride (Ajax Chemicals, Sydney, Australia) in isotonic saline. Control animals received saline only. Animals were studied 5 days after treatment, at which time iron absorption was shown to be maximal. In each experiment, 6 animals were used for each treatment. Iron absorption measurements were carried out on 3 of these, and the other 3 were used in transferrin-binding studies.
Cell Membrane
Preparation
Unless otherwise stated, all intestinal membranes were obtained from the proximal 15 cm (duodenum and proximal jejunum) of the rat small intestine using animals that had been fasted overnight. In some experiments in which larger amounts of membranes were required, the intestines from 2 or 3 animals were pooled. Brush-border membranes were prepared from intestinal scrapings according to the method of Kessler et al. (19). The final membrane pellet was resuspended in binding buffer (50 mM HEPES, 0.1 M KCl, pH 7.4) by passage through a 26G needle. To ensure that membranes were free from endogenous transferrin, they were routinely washed with the chaotrope 3 M KC1 (20) as follows. Membranes were collected by centrifugation then resuspended in 50 mM HEPES buffer (pH 7.4) containing 3 M KCI. Following incubation at 37°C for 30 min. membranes were collected by centrifugation, washed twice in binding buffer, and resuspended in binding buffer as above. Basolateral membranes were prepared from intestinal scrapings by the method of Scalera et al. (21) with minor modifications. The initial homogenization was carried out using a Polytron homogenizer (Kinematica Gmbh, Luzern, Switzerland) rather than in a blender as described (21). In addition, the self-generating density gradients were prepared with a final Percoll (Pharmacia) concentration of 13.7% rather than 10%. The higher concentration was necessary to achieve the desired separation. The rest of the isolation procedure was as described. Chaotrope washing was carried out as detailed above. Erythrocyte-ghosts and placental trophoblast membranes were prepared by the methods of Fairbanks et al. [22) and Smith et al. (231, respectively.
Enzyme
Assays
Sucrase was used as a brush-border marker and was assayed by the method of Dahlqvist (24). Liberated glucose was measured by the hexokinase procedure (Kit 16-UV, Sigma Chemical Co., St. Louis, MO.). Sodium-potassiumstimulated adenosine triphosphatase (Na’, K+,-ATPase] was
578
ANDERSON
GASTROENTEROLOGY
ET AL.
used as a basolateral membrane marker and was measured as ouabain-sensitive ATPase by the method of Rowling and Sepulveda (25). Liberated inorganic phosphate was determined by the method of Ottolenghi (26). Identical results were obtained when K’ was omitted from the incubation medium.
Labeling of Transferrin Transferrin and apotransferrin (50 I.cg) were iodinated by the method of Bolton and Hunter (27) using 1 mCi of Bolton and Hunter reagent (Amersham Australia Ltd.). Unreacted reagent was removed by chromatography on a Sephadex G-25M (PDlO) column.
Binding
Studies
Transferrin binding to membrane preparations was studied using a competitive binding assay. Incubations were carried out in glass tubes (10 x 75 mm) that had been precoated with a 1% solution of bovine serum albumin (BSA) in water to minimize interaction of assay components with the glass. All incubations were for 1 h at 37°C unless otherwise stated. The incubation mixture contained 2.5 pmol of iodinated transferrin (specific activity, 180-270 Ci/mmol] and 50 pg of membrane protein in a total volume of 0.25 ml. For the earlier studies, nonspecific binding was assessed by including 125 pmol of unlabeled transferrin (a 50-fold excess) in a duplicate series of tubes. For the analysis of binding parameters, a range ge of O-125 pmol of unlabeled transferrin was included in each series of tubes. Unless otherwise stated, results are presented as specific binding only. For the specificity experiment the quantities of competing protein are shown in the figure legend. All dilutions were carried out using binding buffer containing 5 mg/ml BSA. Following appropriate incubation, the reaction was terminated by the addition of 10 volumes of ice-cold binding buffer and membranes were collected by vacuum filtration on GVWP membranes (Millipore Pty Ltd., Sydney, Australia). The filters were washed with 15 ml of ice-cold binding buffer, dried, and counted in an LKB Wallac model 1282 gamma counter (Stockholm, Sweden). Data were analyzed by the method of Scatchard (28). Apotransferrin binding was assessed as detailed above, but stringent measures were taken to exclude residual iron. All buffers were passed over a column of Chelex 100, and the BSA used in the experiments was treated to remove any iron (see apotransferrin preparation above). In addition, desferrioxamine was added to the incubation mixture to a final concentration of 10 PM. When membranes were treated with trypsin (Commonwealth Serum Laboratories, Melbourne, Australia], the enzyme was added to a final concentration of 0.013% and the membranes were incubated at 37°C for 30 min. Membranes were collected by centrifugation, washed twice in binding buffer, then used in binding studies as indicated above.
Immunohistology In addition to biochemical analyses, ceptor distribution in the rat small intestine
transferrin rewas examined
Vol. 98, No. 3
by immunohistochemical staining. A murine monoclonal antibody to the rat transferrin receptor (MRC 0X-26) (29) was obtained as ascitic fluid from Dr A. F. Williams (MRC Cellular Immunology Unit, University of Oxford, England). An IgG,, fraction was prepared from the ascites by chromatography on protein A Sepharose (Pharmacia). Tissue for staining was removed from the animal, arranged in OCT embedding compound (Miles Laboratories Australia Pty Ltd., Brisbane, Australia), snap-frozen in liquid nitrogen, and stored at -7O’C until required for sectioning. Frozen sections of 5-6-l.crn thickness were cut, fixed in a mixture of equal volumes of chloroform and acetone at room temperature for 5 min. and stored at -20’C. Sections were routinely stained within 2 days of cutting, although staining intensity was unchanged after storage for at least 7 days. Before staining, sections were thawed and rehydrated in 20 mM phosphate-buffered isotonic saline. The staining procedure involved the sequential addition of the following reagents (a] normal sheep serum (Silenus Laboratories Pty Ltd, Melbourne, Australia] (20 min), (b) monoclonal antibody MRC OX-26 (30 min), (c) biotinylated sheep anti-mouse immunoglobulin (Amersham) (30 min), (d) streptavidin conjugated to horseradish peroxidase (Amersham) (30 min), and (e) 3,3’diaminobenzidine (Sigma] (15 min). All steps were carried out at room temperature for the times indicated. Between incubations the slides were washed three times (each time for 5 min duration) in isotonic saline. Sections were counterstained with Mayer’s hematoxylin and mounted in Glycergel aqueous mounting medium (Dako Corporation, Santa Barbara, Calif.).
Other Methods Hepatic iron content was determined by atomic absorption spectrometry following wet ashing (30). Hematocrits were determined by the microhematocrit procedure using a Clay-Adams Autocrit Centrifuge [New York). Reticulocytes were stained using new methylene blue (31). At least 1000 cells were counted, and the results were expressed as a percentage of total red cell number. Protein concentrations were determined by the BCA method (Pierce, Rockford, Ill.).
Statistical Analysis Data involving two groups were analyzed by Student’s t-test. Data involving multiple groups were analyzed by one-way analysis of variance followed by the NewmanKeul multiple range test to identify the source of any significant differences found (32).
Results Characterization
of Membrane
Vesicles
Electron microscopic examination of the basolateral membranes prepared by the method of Scalera et al. (21) showed a population of closed vesicles of variable size. A typical membrane preparation yielded 1-2 mg of membrane protein for each 15 cm of proximal
small intestine.
Relative
to the initial homoge-
INTESTINAL TRANSFERRIN RECEPTORS
March 1990
nate, ouabain sensitive Na+, Kf-ATPase (a plasma membrane marker] was enriched 18.1 + .%.&fold (n = 61, whereas the brush-border marker sucrase was slightly depleted relative to the homogenate (0.71 + 0.08-fold [n = 611. Brush-border membranes had an electron microscopic appearance similar to that of the basolateral membranes but showed a 17.9 t 2.3fold [n = 6) enrichment in sucrase relative to the homogenate. The enzyme enrichment factors obtained for each membrane type compared favorably with reported values (X%21).
Transferrin
Binding to Intestinal Membranes
Unless otherwise indicated, experiments were carried out using animals receiving the standard rat pellet diet. Incubation of intestinal membranes obtained from animals receiving the rat pellet diet with ‘251-labeled transferrin in the presence or absence of a 50-fold excess of unlabeled transferrin indicated specific binding of transferrin to basolateral membranes but negligible binding to brush-border membranes (Figure 11. Placental trophoblast membranes and erythrocyte membranes were used as positive and negative controls respectively. Placental membranes are a rich source of transferrin receptors and bound a large amount of transferrin, whereas erythrocyte membranes had a low level of specific transferrin binding. It is possible that the residual transferrin binding remaining in the erythrocyte membrane preparation represents membranes derived from immature erythroid cells and lymphoid cells. The binding of transferrin to basolateral membranes was abolished by prior treatment of the membranes with trypsin (Figure 11, confirming the protein nature of the binding site. The data in Figure 2 show the specificity of transferrin binding. Only rat and human transferrin were able to compete with the iodinated rat transferrin for binding sites on the gut membranes. The ability of rat and human transferrin to compete equally well for binding sites on reticulocytes has previously been documented (33), and similar specificity was exhibited by placental membranes (not shown). To determine suitable conditions for the routine analysis of transferrin-binding sites on the gut membranes, the temperature and time dependence of the binding was studied (Figure 3). The lower transferrin binding seen at 4°C than at .z.z~C or 37°C suggests that equilibrium is reached more slowly at the lower temperature. This experiment showed that an incubation time of 60 min at 37’C was suitable for routine studies, and all other experiments were carried out under these conditions. To obtain quantitative information on the number and affinity of transferrin-binding sites on intestinal basolateral membranes, a competitive binding assay and subsequent Scatchard
2 5 ‘, ‘p t
1.6
1.5
2
1
f“E
1.4 :
E 1
0.5
E ,P v s lz .E k L z t 7 l82
579
I*
L
:
*
0.4
0.3 1
-
0.2 ,
_
0.1
BL
0B
Plac
RBC
Membrane Type Figure I. Specific binding of ‘2SI-labeled tramferrin to isolated membranes. The bars (from left to right) represent holotransferrin binding to basolateral (BL) (n = tl), brush-border (BB) (n = 6), placental Cplac) (n = 4 erythrocyte (RBC) (n = 6), and trypsintreated basolateral (Bwryp)) (n = 3) membranes and apotransferrin binding to basolateral membranes (BIJApoTf))(n = 6) Experiments were carried out according to the procedures daecribed in Materials and Methods. SpedSc binding only is rhown. Means and standard deviations are shown. *Significantly different from all other groups (p < 0.001)
analysis were used. A typical curve and its corresponding Scatchard plot are shown in Figure 4. These studies indicated that the binding of transferrin to basolateral membranes was mediated by a single class of high-affinity binding sites. The small amount of transferrin bound to brush-border membranes did not permit effective Scatchard analysis. The high specificity of the binding for transferrin and the protein nature of the binding sites on basolateral membranes provided support for the conclusion that the transferrin receptor was responsible for the observed binding.
Apotransferrin Membranes
Binding to Basolateral
When stringent measures were taken to remove residual iron from proteins, membranes, and reagents, intestinal basolateral membranes bound five times less apotransferrin than diferric transferrin at pH 7.4 (Figure 1). The apotransferrin binding was insufficient to allow reliable Scatchard analysis of the interaction.
580 ANDERSON
-
I
-
Htf
RF
Unlabelled
I
L
* * aRTf
Control
-
-
HLf
Added Non-radiolabellad Transferrin (pmolhbe)
B
-
_
Protein
Transferrin Binding in Different the Small Intestine
Regions of
To determine the levels of transferrin receptors in the main iron-absorbing region of the small intestine relative to other parts of the organ, basolateral membranes were prepared from three regions along the length of the small intestine: O-15 cm [duodenum/ upper jejunum], 35-50 cm (jejunum), and W-105 cm (ileum) distal to the pylorus. The affinity-binding constants of membranes from each of these regions for transferrin were duodenum, 1.07 k 0.21 x lo* M-l
8’
??
10
20 Incubation
30 Tlma
40
.
Ovalb
Figure 2. Specificity of transferrin binding to basolateral membrane vesicles from the intestines of pellet-fed rats. Total binding of 1251-labeledtransferrin is shown in the absence of competing unlabeled protein (control) and in the presence of IO pg of rat transferrin (RTf),human transfer& 0, rat ferritin (RF),human lactoferrin (HLfJor ovalbumin (Ovalb) Means and standard deviations of three separate experiments are shown. “Significantly different from control (p < 0.001).
I
Vol. 98, No. 3
GASTROENTEROLOGY
ET AL.
50
’ ii-120
60
fmlnutms)
Figure 3. Time course and temperature dependence of lzsIlabeled transferrin binding to basolateral membranes from the intestines of pellet-fed rats. Points represent total binding in the presence of 2.5 pmol of labeled transferrin. Duplicate points of a representative experiment are shown.
.
:
I
.
_\j 5
6
7
a
9
10
11
12
13
Transferrin Bound (molxl0”)
Figure 4. [A) Specific binding of 2.5 pmol of ‘ZSI-labeledtransferrin to basolateral membranes from the intestines of pellet-fed rats in the presence of increasing concentrations of unlabeled transferrin. Duplicate points are shown. (B) Scatchard plot of binding data derived from the means of the duplicate points shown in A.
(n = 3); jejunum, 1.10 + 0.07 x 10s M-l [n = 3); and ileum, 1.39 + 0.33 x lo* M-’ (n = 3) (no statistically significant differences between groups), while the corresponding B,,, values were 632 r 59 (n = 31, 454 + 41 (n = 31, and 412 + 78 (n = 3) fmol/mg of membrane protein (jejunal and ileal values significantly different from the duodenal value, p -C0.01) These data indicate that receptor affinity was the same at each site of the intestine, whereas receptor number was significantly greater in the duodenum/proximal jejunum. Relationship Absorption
Between Iron Stores and Iron
Variations in body iron stores were produced in five groups of animals either by feeding iron-deficient or iron-supplemented diets or by parenteral iron administration. The iron-replete diet was intended to replace the rat pellet diet in this series of experiments because it could be prepared with a known composition and would obviate any batch differences associated with the rat pellet diet. However, the rat pellet diet consistently provided animals with iron stores intermediate between those on the iron-deficient and iron-replete diets, and those animals proved a useful additional group. Hepatic iron concentration was used as an indicator of iron status, and the results in Table I
March 1990
INTESTINAL TRANSFERRIN
RECEF’TORS
581
Table 1. Iron Status and iron Absorption in Experimental Groups* Hepatic Iron concentration (pmol/g dry wt)
Weight Group Iron-deficient Rat pellet diet Iron-replete Carbonyl iron-supplemented Iron dextran-supplemented
(gl 265 + 33 (141 290 + 37 (151 304 + 34
2.1 + 1.2abc (14) 4.3 + l.OCd
Hematocrit (%I
Reticulocyte count (%I
Iron absorption 1%)
8.7 + 1 6”kd
71.6 f 6.4”‘“d
(5; 1.0 * 0.4
(81 10.1 k 2.8e’
21.9 k 3.6”b”” (5) 46.6 + 1.3a (51 45.7 * 1.3b
I51 29.7 * 9.70’6 (51
(191
(19)
(51
(51
256 -+34” (14) 309 + 72a
266 f 60aceg (10) 227 f 3gMC
(181 *Results represent mean f SD. Number of determinations
(18)
47.2 r 4.1’ (5) 46.5 + 1.8d (81
(51 1.1 r 0.6” 0.7 -c 0.5” (51 0.6 z+0.2d (51
11.2 * 1.9"eh (51
2.6 k l.l”fi (51 5.7 k 3.6dK (51
isshown in parentheses.
Matchingletters in each column represent significant differences between groups (p 5 0.05).
clearly show that the methods used effectively produced a wide range of body iron stores. The absorption of radioactive iron was carried out in parallel in separate groups of animals and was inversely proportional to the iron stores (Table 1). In the iron-deficient group, almost three quarters of the tracer dose was absorbed, whereas
Transferrin Receptor Distribution and Regulation in the Rat Small Intestine Effect of Iron Stores and Erythropoiesis GREGORY J. ANDERSON, LAWRIE W. POWELL, and JUNE W. HALLIDAY Department of Medicine, University of Queensland, Brisbane,Queensland, Australia
A combination of biochemical quantitation and immunohistochemistry has been used to examine in detail transferrin receptor distribution and expression in the rat small intestine and its relationship to iron absorption. Receptor numbers were quantitated by transferrin binding to preparations of basolateral or brush-border membranes. Receptors were demonstrated on the basolateral membranes of the gut cells, but not on the brush-border fraction. Apotransferrin demonstrated little binding to basolateral membranes at physiological pH. Dietary or parenteral iron loading of animals produced a significant decline in transferrin binding, whereas binding was increased in iron deficiency. These data were confirmed by immunohistochemical studies using a monoclonal antibody to the transferrin receptor. When iron absorption was increased threefold following acute hemolysis and without a decrease in body iron stores, there was no change in transferrin receptor number. These data indicate that intestinal transferrin receptors may be regulated by body iron stores but suggest that they are not directly involved in iron absorption.
he intestinal epithelial cell is one of the few cell types that transport iron vectorially. Iron must be taken into the cell across the brush-border membrane, transported through the cytoplasmic compartment, and released across the basal and/or lateral membranes into the extravascular fluid (1). Some of the details of brush-border uptake have been determined (2-6) and several intracellular iron-binding proteins have been identified (Y-10) but very little is known about the handling of iron at the serosal side of the enterocyte. A number of recent histochemical studies have shown transferrin receptors on the basolateral mem-
T
branes of intestinal epithelial cells (11-13). These have led to speculation on the role of the transferrin receptor in the gut mucosa and the role of transferrin in iron absorption (14). In addition, there is a proposal that transferrin is secreted into the intestinal lumen, binds iron, and is reabsorbed by a receptor-mediated process (15). It is not yet known whether enterocytes internalize transferrin by endocytosis, and there are few biochemical data on intestinal transferrin receptors, but because the receptor-mediated uptake of transferrin provides a major pathway of iron uptake by cells (16) the transferrin receptor may play an important role in the response of the intestinal cell to physiological stimuli. Most previous studies of intestinal transferrin receptors either have used semiquantitative immunohistochemical techniques or have been incidental observations made as parts of other studies. The purpose of the present report is to provide quantitative biochemical data on the distribution and numbers of intestinal transferrin receptors in relation to iron absorption. Using the rat as a model system we have been able to carry out a systematic study of the effects of body iron stores and erythropoiesis on intestinal receptor levels. These data suggest that intestinal transferrin receptor expression is related to iron stores, but that the receptor is not directly involved in iron absorption.
Materials and Methods Animals and Diets Adult male Sprague-Dawley rats were used for all experiments. Animals were divided into five groups:
Abbreviation used in this paper: BSA,bovine serum albumin. 0 1990 by the American Gashoenterological Association 0016-5065/90/$3.00
INTESTINAL TRANSFERRIN RECEPTORS 577
March 1990
1. Rat pellet-fed 2.
3.
4.
5.
animals received standard rodent pellet diet (determined iron content, 47 mg/kg wet wt). Iron-replete animals received the iron-deficient diet described by Valberg et al. (17) supplemented with ferrous ammonium sulfate (0.75 g/kg dry wt) (determined iron content, 159 mg/kg wet wt). Iron-deficient animals received the iron-deficient diet described by Valberg et al. (17) (determined iron content, 3 mg/kg wet wt). In addition, these animals were allowed access to deionized water only and were housed in all-plastic cages with glass water bottles. These conditions were essential for iron deficiency to be achieved and maintained. Carbonyl iron-supplemented animals received the irondeficient diet supplemented with 2.5% wt/wt (dry wt) carbonyl iron (GAF Corp., New York, N.Y.) (determined iron content, 14.9 g/kg wet wt). This diet produces hepatic parenchymal cell iron loading (18). Iron dextran-supplemented animals were maintained on the rat pellet diet but received intramuscular injections of 50 mg of iron dextran (Imferon; Fisons Pty Ltd., Sydney, Australia) in each of five successive weeks. This treatment produced iron-loaded animals in which iron was predominantly localized in the reticuloendothelial system.
The iron content of each diet was determined by atomic absorption spectrometry. Animals began receiving the irondeficient diet from the time of weaning, whereas all other treatments were started when the animals weighed 150 g. Animals in which reticulocytosis was induced [see below) received the standard rat pellet diet. All experiments described in this study were approved by the Animal Experimentation Ethics Committee of the university.
Transferrin
and Apotransferrin
Preparation
Transferrin was prepared from the pooled sera of normal rats by sequential ion exchange (Whatman DE-52 cellulose: Whatman Ltd., Maidstone, Kent, England) and size exclusion (Sephacryl S-200; Pharmacia [Australia] Pty. Ltd., Sydney, Australia) chromatography of an ammonium sulfate fraction precipitating between 40% and 70% saturation (7). Apotransferrin was prepared by dialyzing holotransferrin overnight against 50 mM sodium acetate (pH 5.0) containing 0.25 mM desferrioxamine (as mesylate [Desferal; Ciba Geigy Pty Ltd., Sydney, Australia]), followed by incubation with an equal volume of Chelex 100 (Bio-Rad Laboratories, Sydney, Australia) for 1 h at room temperature. The Chelex was removed by low-speed centrifugation.
Iron Absorption
Measurements
Iron absorption was measured by whole-body counting following administration of a tracer dose of 5gFe. For each 250-~1 dose, 0.5 &i of 59Fe ferric chloride (Amersham Australia Ltd., Sydney, Australia) was added to a solution of ferrous sulfate in 10 mM HCI to give a final iron concentration of 200 wg/ml. The dose was administered intragastritally to unanesthetized rats that had been fasted overnight. Whole-body radioactivity was determined 2-3 h and 7 days
after iron administration using a ARMAC large-volume gamma counter (Packard Instrument Company, Downers Grove, Ill.). Absorption was calculated as the percentage of iron remaining in the body 7 days after correction for radionuclide decay.
Induction
of Increased
Erythropoiesis
An increase in erythropoiesis followed reticulocytosis induced in 200-g rats by the subcutaneous injection of a single lo-mg dose of neutralized phenylhydrazine hydrochloride (Ajax Chemicals, Sydney, Australia) in isotonic saline. Control animals received saline only. Animals were studied 5 days after treatment, at which time iron absorption was shown to be maximal. In each experiment, 6 animals were used for each treatment. Iron absorption measurements were carried out on 3 of these, and the other 3 were used in transferrin-binding studies.
Cell Membrane
Preparation
Unless otherwise stated, all intestinal membranes were obtained from the proximal 15 cm (duodenum and proximal jejunum) of the rat small intestine using animals that had been fasted overnight. In some experiments in which larger amounts of membranes were required, the intestines from 2 or 3 animals were pooled. Brush-border membranes were prepared from intestinal scrapings according to the method of Kessler et al. (19). The final membrane pellet was resuspended in binding buffer (50 mM HEPES, 0.1 M KCl, pH 7.4) by passage through a 26G needle. To ensure that membranes were free from endogenous transferrin, they were routinely washed with the chaotrope 3 M KC1 (20) as follows. Membranes were collected by centrifugation then resuspended in 50 mM HEPES buffer (pH 7.4) containing 3 M KCI. Following incubation at 37°C for 30 min. membranes were collected by centrifugation, washed twice in binding buffer, and resuspended in binding buffer as above. Basolateral membranes were prepared from intestinal scrapings by the method of Scalera et al. (21) with minor modifications. The initial homogenization was carried out using a Polytron homogenizer (Kinematica Gmbh, Luzern, Switzerland) rather than in a blender as described (21). In addition, the self-generating density gradients were prepared with a final Percoll (Pharmacia) concentration of 13.7% rather than 10%. The higher concentration was necessary to achieve the desired separation. The rest of the isolation procedure was as described. Chaotrope washing was carried out as detailed above. Erythrocyte-ghosts and placental trophoblast membranes were prepared by the methods of Fairbanks et al. [22) and Smith et al. (231, respectively.
Enzyme
Assays
Sucrase was used as a brush-border marker and was assayed by the method of Dahlqvist (24). Liberated glucose was measured by the hexokinase procedure (Kit 16-UV, Sigma Chemical Co., St. Louis, MO.). Sodium-potassiumstimulated adenosine triphosphatase (Na’, K+,-ATPase] was
578
ANDERSON
GASTROENTEROLOGY
ET AL.
used as a basolateral membrane marker and was measured as ouabain-sensitive ATPase by the method of Rowling and Sepulveda (25). Liberated inorganic phosphate was determined by the method of Ottolenghi (26). Identical results were obtained when K’ was omitted from the incubation medium.
Labeling of Transferrin Transferrin and apotransferrin (50 I.cg) were iodinated by the method of Bolton and Hunter (27) using 1 mCi of Bolton and Hunter reagent (Amersham Australia Ltd.). Unreacted reagent was removed by chromatography on a Sephadex G-25M (PDlO) column.
Binding
Studies
Transferrin binding to membrane preparations was studied using a competitive binding assay. Incubations were carried out in glass tubes (10 x 75 mm) that had been precoated with a 1% solution of bovine serum albumin (BSA) in water to minimize interaction of assay components with the glass. All incubations were for 1 h at 37°C unless otherwise stated. The incubation mixture contained 2.5 pmol of iodinated transferrin (specific activity, 180-270 Ci/mmol] and 50 pg of membrane protein in a total volume of 0.25 ml. For the earlier studies, nonspecific binding was assessed by including 125 pmol of unlabeled transferrin (a 50-fold excess) in a duplicate series of tubes. For the analysis of binding parameters, a range ge of O-125 pmol of unlabeled transferrin was included in each series of tubes. Unless otherwise stated, results are presented as specific binding only. For the specificity experiment the quantities of competing protein are shown in the figure legend. All dilutions were carried out using binding buffer containing 5 mg/ml BSA. Following appropriate incubation, the reaction was terminated by the addition of 10 volumes of ice-cold binding buffer and membranes were collected by vacuum filtration on GVWP membranes (Millipore Pty Ltd., Sydney, Australia). The filters were washed with 15 ml of ice-cold binding buffer, dried, and counted in an LKB Wallac model 1282 gamma counter (Stockholm, Sweden). Data were analyzed by the method of Scatchard (28). Apotransferrin binding was assessed as detailed above, but stringent measures were taken to exclude residual iron. All buffers were passed over a column of Chelex 100, and the BSA used in the experiments was treated to remove any iron (see apotransferrin preparation above). In addition, desferrioxamine was added to the incubation mixture to a final concentration of 10 PM. When membranes were treated with trypsin (Commonwealth Serum Laboratories, Melbourne, Australia], the enzyme was added to a final concentration of 0.013% and the membranes were incubated at 37°C for 30 min. Membranes were collected by centrifugation, washed twice in binding buffer, then used in binding studies as indicated above.
Immunohistology In addition to biochemical analyses, ceptor distribution in the rat small intestine
transferrin rewas examined
Vol. 98, No. 3
by immunohistochemical staining. A murine monoclonal antibody to the rat transferrin receptor (MRC 0X-26) (29) was obtained as ascitic fluid from Dr A. F. Williams (MRC Cellular Immunology Unit, University of Oxford, England). An IgG,, fraction was prepared from the ascites by chromatography on protein A Sepharose (Pharmacia). Tissue for staining was removed from the animal, arranged in OCT embedding compound (Miles Laboratories Australia Pty Ltd., Brisbane, Australia), snap-frozen in liquid nitrogen, and stored at -7O’C until required for sectioning. Frozen sections of 5-6-l.crn thickness were cut, fixed in a mixture of equal volumes of chloroform and acetone at room temperature for 5 min. and stored at -20’C. Sections were routinely stained within 2 days of cutting, although staining intensity was unchanged after storage for at least 7 days. Before staining, sections were thawed and rehydrated in 20 mM phosphate-buffered isotonic saline. The staining procedure involved the sequential addition of the following reagents (a] normal sheep serum (Silenus Laboratories Pty Ltd, Melbourne, Australia] (20 min), (b) monoclonal antibody MRC OX-26 (30 min), (c) biotinylated sheep anti-mouse immunoglobulin (Amersham) (30 min), (d) streptavidin conjugated to horseradish peroxidase (Amersham) (30 min), and (e) 3,3’diaminobenzidine (Sigma] (15 min). All steps were carried out at room temperature for the times indicated. Between incubations the slides were washed three times (each time for 5 min duration) in isotonic saline. Sections were counterstained with Mayer’s hematoxylin and mounted in Glycergel aqueous mounting medium (Dako Corporation, Santa Barbara, Calif.).
Other Methods Hepatic iron content was determined by atomic absorption spectrometry following wet ashing (30). Hematocrits were determined by the microhematocrit procedure using a Clay-Adams Autocrit Centrifuge [New York). Reticulocytes were stained using new methylene blue (31). At least 1000 cells were counted, and the results were expressed as a percentage of total red cell number. Protein concentrations were determined by the BCA method (Pierce, Rockford, Ill.).
Statistical Analysis Data involving two groups were analyzed by Student’s t-test. Data involving multiple groups were analyzed by one-way analysis of variance followed by the NewmanKeul multiple range test to identify the source of any significant differences found (32).
Results Characterization
of Membrane
Vesicles
Electron microscopic examination of the basolateral membranes prepared by the method of Scalera et al. (21) showed a population of closed vesicles of variable size. A typical membrane preparation yielded 1-2 mg of membrane protein for each 15 cm of proximal
small intestine.
Relative
to the initial homoge-
INTESTINAL TRANSFERRIN RECEPTORS
March 1990
nate, ouabain sensitive Na+, Kf-ATPase (a plasma membrane marker] was enriched 18.1 + .%.&fold (n = 61, whereas the brush-border marker sucrase was slightly depleted relative to the homogenate (0.71 + 0.08-fold [n = 611. Brush-border membranes had an electron microscopic appearance similar to that of the basolateral membranes but showed a 17.9 t 2.3fold [n = 6) enrichment in sucrase relative to the homogenate. The enzyme enrichment factors obtained for each membrane type compared favorably with reported values (X%21).
Transferrin
Binding to Intestinal Membranes
Unless otherwise indicated, experiments were carried out using animals receiving the standard rat pellet diet. Incubation of intestinal membranes obtained from animals receiving the rat pellet diet with ‘251-labeled transferrin in the presence or absence of a 50-fold excess of unlabeled transferrin indicated specific binding of transferrin to basolateral membranes but negligible binding to brush-border membranes (Figure 11. Placental trophoblast membranes and erythrocyte membranes were used as positive and negative controls respectively. Placental membranes are a rich source of transferrin receptors and bound a large amount of transferrin, whereas erythrocyte membranes had a low level of specific transferrin binding. It is possible that the residual transferrin binding remaining in the erythrocyte membrane preparation represents membranes derived from immature erythroid cells and lymphoid cells. The binding of transferrin to basolateral membranes was abolished by prior treatment of the membranes with trypsin (Figure 11, confirming the protein nature of the binding site. The data in Figure 2 show the specificity of transferrin binding. Only rat and human transferrin were able to compete with the iodinated rat transferrin for binding sites on the gut membranes. The ability of rat and human transferrin to compete equally well for binding sites on reticulocytes has previously been documented (33), and similar specificity was exhibited by placental membranes (not shown). To determine suitable conditions for the routine analysis of transferrin-binding sites on the gut membranes, the temperature and time dependence of the binding was studied (Figure 3). The lower transferrin binding seen at 4°C than at .z.z~C or 37°C suggests that equilibrium is reached more slowly at the lower temperature. This experiment showed that an incubation time of 60 min at 37’C was suitable for routine studies, and all other experiments were carried out under these conditions. To obtain quantitative information on the number and affinity of transferrin-binding sites on intestinal basolateral membranes, a competitive binding assay and subsequent Scatchard
2 5 ‘, ‘p t
1.6
1.5
2
1
f“E
1.4 :
E 1
0.5
E ,P v s lz .E k L z t 7 l82
579
I*
L
:
*
0.4
0.3 1
-
0.2 ,
_
0.1
BL
0B
Plac
RBC
Membrane Type Figure I. Specific binding of ‘2SI-labeled tramferrin to isolated membranes. The bars (from left to right) represent holotransferrin binding to basolateral (BL) (n = tl), brush-border (BB) (n = 6), placental Cplac) (n = 4 erythrocyte (RBC) (n = 6), and trypsintreated basolateral (Bwryp)) (n = 3) membranes and apotransferrin binding to basolateral membranes (BIJApoTf))(n = 6) Experiments were carried out according to the procedures daecribed in Materials and Methods. SpedSc binding only is rhown. Means and standard deviations are shown. *Significantly different from all other groups (p < 0.001)
analysis were used. A typical curve and its corresponding Scatchard plot are shown in Figure 4. These studies indicated that the binding of transferrin to basolateral membranes was mediated by a single class of high-affinity binding sites. The small amount of transferrin bound to brush-border membranes did not permit effective Scatchard analysis. The high specificity of the binding for transferrin and the protein nature of the binding sites on basolateral membranes provided support for the conclusion that the transferrin receptor was responsible for the observed binding.
Apotransferrin Membranes
Binding to Basolateral
When stringent measures were taken to remove residual iron from proteins, membranes, and reagents, intestinal basolateral membranes bound five times less apotransferrin than diferric transferrin at pH 7.4 (Figure 1). The apotransferrin binding was insufficient to allow reliable Scatchard analysis of the interaction.
580 ANDERSON
-
I
-
Htf
RF
Unlabelled
I
L
* * aRTf
Control
-
-
HLf
Added Non-radiolabellad Transferrin (pmolhbe)
B
-
_
Protein
Transferrin Binding in Different the Small Intestine
Regions of
To determine the levels of transferrin receptors in the main iron-absorbing region of the small intestine relative to other parts of the organ, basolateral membranes were prepared from three regions along the length of the small intestine: O-15 cm [duodenum/ upper jejunum], 35-50 cm (jejunum), and W-105 cm (ileum) distal to the pylorus. The affinity-binding constants of membranes from each of these regions for transferrin were duodenum, 1.07 k 0.21 x lo* M-l
8’
??
10
20 Incubation
30 Tlma
40
.
Ovalb
Figure 2. Specificity of transferrin binding to basolateral membrane vesicles from the intestines of pellet-fed rats. Total binding of 1251-labeledtransferrin is shown in the absence of competing unlabeled protein (control) and in the presence of IO pg of rat transferrin (RTf),human transfer& 0, rat ferritin (RF),human lactoferrin (HLfJor ovalbumin (Ovalb) Means and standard deviations of three separate experiments are shown. “Significantly different from control (p < 0.001).
I
Vol. 98, No. 3
GASTROENTEROLOGY
ET AL.
50
’ ii-120
60
fmlnutms)
Figure 3. Time course and temperature dependence of lzsIlabeled transferrin binding to basolateral membranes from the intestines of pellet-fed rats. Points represent total binding in the presence of 2.5 pmol of labeled transferrin. Duplicate points of a representative experiment are shown.
.
:
I
.
_\j 5
6
7
a
9
10
11
12
13
Transferrin Bound (molxl0”)
Figure 4. [A) Specific binding of 2.5 pmol of ‘ZSI-labeledtransferrin to basolateral membranes from the intestines of pellet-fed rats in the presence of increasing concentrations of unlabeled transferrin. Duplicate points are shown. (B) Scatchard plot of binding data derived from the means of the duplicate points shown in A.
(n = 3); jejunum, 1.10 + 0.07 x 10s M-l [n = 3); and ileum, 1.39 + 0.33 x lo* M-’ (n = 3) (no statistically significant differences between groups), while the corresponding B,,, values were 632 r 59 (n = 31, 454 + 41 (n = 31, and 412 + 78 (n = 3) fmol/mg of membrane protein (jejunal and ileal values significantly different from the duodenal value, p -C0.01) These data indicate that receptor affinity was the same at each site of the intestine, whereas receptor number was significantly greater in the duodenum/proximal jejunum. Relationship Absorption
Between Iron Stores and Iron
Variations in body iron stores were produced in five groups of animals either by feeding iron-deficient or iron-supplemented diets or by parenteral iron administration. The iron-replete diet was intended to replace the rat pellet diet in this series of experiments because it could be prepared with a known composition and would obviate any batch differences associated with the rat pellet diet. However, the rat pellet diet consistently provided animals with iron stores intermediate between those on the iron-deficient and iron-replete diets, and those animals proved a useful additional group. Hepatic iron concentration was used as an indicator of iron status, and the results in Table I
March 1990
INTESTINAL TRANSFERRIN
RECEF’TORS
581
Table 1. Iron Status and iron Absorption in Experimental Groups* Hepatic Iron concentration (pmol/g dry wt)
Weight Group Iron-deficient Rat pellet diet Iron-replete Carbonyl iron-supplemented Iron dextran-supplemented
(gl 265 + 33 (141 290 + 37 (151 304 + 34
2.1 + 1.2abc (14) 4.3 + l.OCd
Hematocrit (%I
Reticulocyte count (%I
Iron absorption 1%)
8.7 + 1 6”kd
71.6 f 6.4”‘“d
(5; 1.0 * 0.4
(81 10.1 k 2.8e’
21.9 k 3.6”b”” (5) 46.6 + 1.3a (51 45.7 * 1.3b
I51 29.7 * 9.70’6 (51
(191
(19)
(51
(51
256 -+34” (14) 309 + 72a
266 f 60aceg (10) 227 f 3gMC
(181 *Results represent mean f SD. Number of determinations
(18)
47.2 r 4.1’ (5) 46.5 + 1.8d (81
(51 1.1 r 0.6” 0.7 -c 0.5” (51 0.6 z+0.2d (51
11.2 * 1.9"eh (51
2.6 k l.l”fi (51 5.7 k 3.6dK (51
isshown in parentheses.
Matchingletters in each column represent significant differences between groups (p 5 0.05).
clearly show that the methods used effectively produced a wide range of body iron stores. The absorption of radioactive iron was carried out in parallel in separate groups of animals and was inversely proportional to the iron stores (Table 1). In the iron-deficient group, almost three quarters of the tracer dose was absorbed, whereas
