JOURNAL OF CELLULAR PHYSlOLOGY 152:102-110 (1992)
Release of Iron by Human Retinal Pigment Epithelial Cells RICHARD C. H U N T * AND ALBERTA A. DAVIS Department of Microbiology, Univeriity of South Carolina Medicdl School, Columbia, South Carolina 29208 Retinal pigment epithelial cells, which form one aspect of the blood-retinal barrier, take up iron in association with transferrin by a typical receptor-mediated mechanism (Hunt et al., 1989. J. Cell Sci. 92:655-666). This iron is dissociated from transferrin in a low pH environment and uptake i s sensitive to agents that inhibit endosomal acidification. The dissociated iron enters the cytoplasm as a low molecular weight (< 10 kD) component and subsequently binds to ferritin. No evidence for recycling of iron in association with transferrin was found. Nevertheless, much of the iron that is taken up is recycled to the extracellular medium, primarily from the low molecular weight pool. This release of iron is not sensitive to inhibitors of energy production or of vesicular acidification but is increased u p to a maximum of about 40% of the total 55Feincorporated when cells are incubated with serum or the medium is changed. When a short loading time for ""Fe from 55Fe-transferrin is used (i.e., when the low molecular weight pool i s proportionately larger), a much larger fraction of the cell-associated radiolabel is released than when longer loading times are used. The data suggest that a releasable intracellular iron pool is in equilibrium with the externalized material. The released iron may be separated into a high and a low molecular weight component. The former i s similar on polyacrylamide gel electrophoresis to ferritin although it cannot be immune precipitated by anti-ferritin antibodies. The low molecular weight 55Fewhich i s heterogeneous in nature can be bound by external apo-transferrin and may represent a form that can be taken up by cells beyond the blood-retinal barrier. o 1992 Wiiev-Liss, Inc.
Free diffusion of blood-borne substances into the neu- ceptor cells since although they do not divide, they ral retina is obstructed by barrier layers of cells that carry out extensive membrane production in order to give the eye its immune privileged properties. The compensate for the shedding of the disk membranes blood-retinal barrier (BRB) comprises the tight junc- that contain the photoreceptor pigment. Thus, it is not tions of the endothelial cells of the retinal capillaries surprising that lack of iron has been suggested t o lead and of the retinal pigment epithelium (Pedersen, 1979; t o retinal dysfunction (Lakhanpal et al., 1984). Peyman and Bok, 1972; Smith and Rudt, 1975) and any The mechanism by which iron is taken up at the nutrient, hormone, or growth factor that must enter the basolateral surface of RPE cells, transported to their neural retina must first cross these cells. Thus selectiv- apical surface, and released for onward migration to the ity of access to, for example, the photoreceptor cells is neural retina is unknown though transferrin receptors determined by whether BRB cells can bind and trans- (TfRs)of the type found on almost all other cells in the port that substance. body are located on the basolateral surfaces of RPE The retinal pigment epithelium (RPE)is a monolayer (Hunt et al., 1989) and other epithelial cells (Fuller and of polarized cells containing melanin and lipofuscin Simons, 1986). After binding and internalization of Tf (Clark, 1986). Their basolateral surface is in contact a t the basolateral surfaces of RPE cells, one of two with the Bruch's membrane and the circulation via the mechanisms for onward transport is most likely. In one fenestrated capillaries of the choroid while their apical model (Fig. lA), Tf is internalized in vesicles that misurfaces embrace the outer segments of the photorecep- grate to the apical surface where it is released and tors. Intercellular diffusion is blocked by tight junc- binds to TfRs on photoreceptor and other cells. In a tional complexes that girdle the cell (Cunha-Vaz, second more complicated model (Fig. lB), Tf loses its 1979). iron in the low pH endosomes of RPE cells and apo-Tf is Iron, carried in the circulation in association with the recycled to the circulation. The iron enters the cytocarrier protein, transferrin (TO,is an obligatory growth factor for all cells (Barnes and Sato, 19801, being involved in a variety of processes that require iron-containing enzymes. Among the latter are fatty acid desaturases (Shichi, 1969) that are important in membrane Received J u l y 30, 1991; accepted February 10, 1992. biogenesis. These enzymes are important to photore- *To whom reprint requests/correspondence should be addressed. 0 1992 WI1,EY-LISS, INC.
PIGMENT EPITHELIAL CELL IRON METABOLISM
apical
e
troloacetic acid and sodium bicarbonate (Klausner et al., 1983b). Free Na’”1 or “FeC1, were removed by Sephadex G25 gel filtration.
e
Q
A
b
B
Q t
Tfe
basolateral
103
apoTf
0 0
Fig. 1. Transcytosis of iron by RPE cells. Model A (left):Diferric Tf (filled circles) binds to a TfR on the hasolateral surface of a n RPE cell, The receptormf complex is internalized and transported to the apical surface where the Tf dissociates for onward migration to the cells of the neural retina. Model B (right):Diferric Tf hinds to a TfR on the hasolateral surface of the cell and is internalized into a low pH endosome where iron (slashed circles) dissociates from the Tf. The apo-Tf (open circles) recycles to the basolateral surface of the cells where it is released from the TfR. The released iron is transported across the membrane of the endosome, probably hound to a hypothetical carrier and enters the cytoplasm. Here it dissociates from the carrier and binds to cytoplasmic apo-ferritin (squares). The cytoplasmic iron pool, either in association with ferritin or the hypothetical carrier molecule, is the source of iron that is secreted from the apical surface of the cell and which binds to apo-Tf in the interphotoreceptor matrix. This differric Tf (filled circles) then hinds to TfRs on cells of the neural retina. The cytoplasmic iron that is not complexed to ferritin and that is secreted into the interphotoreceptor matrix may be bound by the same or different carrier molecules.
plasm where it is complexed to ferritin or some other cytoplasmic chelating agent and this cytoplasmic pool is the source of iron that is secreted at the apical surface for the use of cells beyond the BRB. In this paper, evidence for the latter model is presented.
MATERIALS AND METHODS Culture of RPE cells RPE cells from human donor eyes were grown in Ham’s F10 medium as previously described (Hunt et al., 1989). Preparation of radiolabeled Tf Diferric Tf (ICN Biomedical) was dissolved in phosphate buffered saline (PBS) and iodinated using chloramine T and Na1251(Amersham) as described (Hunt et al., 1984). Apo-Tf (Boehringer-Mannheim) was labeled with 55FeC1, (Amersham) in the presence of ni-
Abbreviations BRB. blood-retinal barrier: BSA. bovine serum albumin; PAGE, polvacrylamide gel electrophoresis; PBS, phosphate buffered saline; RPE, retinal pigment epithelial; SDS, sodium dodecyl sulfate; Tf, transferrin; lYR, transferrin receptor
Non-denaturing polyacrylamide gel electrophoresis 55 Fe-labeled material was analyzed by polyacrylamide gel electrophoresis (PAGE) using the buffer system of Laemmli (Laemmli, 1970) but omitting sodium dodecyl sulfate (SDS) from the electrode buffers and the gel. Labeled protein was applied directly to the gel without SDS or dithiothreitol. After electrophoresis, the gel was fixed in methanol/water (1:l) for 30 minutes and incubated in salicylic acid solution for 20 minutes (Hunt et al., 1984). The gel was then dried and autoradiographed. Incubation of RPE cells The conditions for individual experiments are described in the legends to the Figures. In all cases, binding and uptake of radiolabeled Tf were carried out in Dulbecco’s modified Eagle’s (DME) medium (Gibco) supplemented with penicillin, streptomycin, and bovine serum albumin (BSA) (1 mg/ml). Non-specific binding and uptake were measured in the presence of 100 fold excess unlabeled Tf. The non-specific binding was less than 10% of the specific binding. Removal of surface-boundTf After incubation of cells with radiolabeled Tf, the cells were washed three times at 4°Cwith PBS and then incubated a t 4°C with 0.2 M glycine HC1 (pH 3.5) containing 100 mM NaCl for 5 minutes. The cells were washed again in PBS. Release of cytoplasmic components 0.5 ml PBS was added to a monolayer of washed cells which were then frozen at -20°C. After thawing, the cells were scraped from the bottle with a rubber policeman and centrifuged at 15,000 rpm (16,OOOg) on a Beckman Microfuge for 5 minutes to yield a crude soluble fraction. The small molecular weight molecules were separated from the remainder of the fraction by centrifugation through a Centricon 30 or 10 kD dialysis filter (Amicon). Separation of low and high molecular weight material excreted from RPE cells Medium into which cells had recycled radiolabeled material was centrifuged through Centricon 30 kD or 10 kD cut off filters (Amicon). RESULTS Uptake of Tf and iron by RPE cells The protein moiety of Tf was labeled with 1251and incubated with cultured human RPE cells for various periods of time at 37°C. The cell-associated Tf was determined to be either surface or internal by removing the former in a low pH wash. Non-specific binding was assessed in parallel incubations in the presence of excess unlabeled Tf and was found to be less than 10% of the specific uptake. All non-specific binding appeared to be cell surface (released by low pH) with none internalized. Figure 2A shows that both surface and internal
HUNT AND DAVIS
104 300-
*O0
r
/
700
-
240
=
'
a,
180
-
600
500
0
6
0
6
0
20
40
60
60
120
180
240
100
80
Time (minutes) 0
240
180
Fig. 3. Uptake of 55Fe from 55Fe-Tf by RPE cells. RPE cells were incubated with 55Fe-Tfor "'I-Tf for various periods of time and were treated in a manner similar to those in Figure 2. ( 0 ) 1251-Tf. (A) 55Fe-Tf.( 0 )55Fe-Tfin the presence of 10 pM monensin.
1 I
120
0
t 0
partment was shown by the use of monensin which totally abolished iron accumulation by RPE cells (Fig. 3, solid symbols).
20
40
60
80
100
Time (minutes) Fig. 2. Uptake of I2'II-Tf by RPE cells. A: RPE cells in sub-confluent monolayer culture were incubated with 1251-Tfin serum-free medium supplemented with BSA (1 mg/ml) for various periods of time. After washing the cells at 4"C, they were incubated with low pH buffered saline (0.2M glycine HCI, pH 3.5 containing lO0nM NaCl) to remove surface-associated Tf. The cells were then dissolved in 0.1 M NaOH. Radioactivity associated with the acid wash and the cells was measured by T counting. ( 0 ) Internal "'I. ( 0 ) Surface lZ51.B: The effect of monensin (10KM)on the uptake of '""I-Tfwas measured in a similar experiment. In all cases total binding is corrected for non-specific binding by subtracting the binding in the presence of 100 fold excess unlabeled Tf. (0)Internal Iz5I.( 0 ) Surface lZ5I.
Tf were accumulated by the cells and that accumulation reached a plateau level by about an hour. The saturation of uptake at the plateau level is similar to observations with other cells in which apo-Tf is recycled to the cells' environment (Hanover and Dickson, 1985). At the plateau level there is equilibrium between uptake and recycling. To determine whether the apparent recycling of apo-Tf by RPE cells occurs via a low pH endosomal compartment, cells were incubated with ?-Tf in the presence of monensin which inhibits endosomal acidification (Fig. 2B). In this case, surface binding and cellular accumulation are both reduced by about 60%,resulting probably from the inhibition of TfR recycling in the absence of acidification. In contrast to lz5I-Tf, the cellular accumulation of "Fe-Tf did not reach a plateau (at least up to 4 hours) (Fig. 3) although, as expected, surface binding did exhibit a plateau value (data not shown). Thus, iron must be removed from the Tf for intracellular accumulation. That this was due to passage through an acidic com-
Transit of 5"Fethrough a low molecular weight pool In order to assess the nature of the molecules to which 55Feis bound after endocytosis of "Fe-labeled Tf, cells were incubated for various periods of time (1 minute to 2 hours) with 55Fe-Tf, washed in low pH solution to remove any surface bound Tf, and then frozen and thawed to release cytoplasmic components. The cytoplasmic fraction was then passed through a membrane filter with a molecular weight cut off of 30 kD to separate small and large iron-binding molecules, At early time points, the majority of the Fe is associated with a molecule that passes through the filter. The pool of this small molecular weight cytoplasmic component is rapidly saturated with "Fe (Fig. 4A). The proportion of the label that is in this fraction falls rapidly as iron is accumulated in a > 30 kD cytoplasmic component that does not pass through the filter (Fig. 4B). This material comigrates with ferritin on non-denaturing PAGE (data not shown). The "Fe-labeled small molecular weight material associated with the cells is not simply the adherence of an impurity in the 55Fe-Tfpreparation since i) it does not occur a t 4"C, ii) it is abolished by excess unlabeled Tf, and iii) it is not removed by a low pH incubation. Release of iron to the medium by RPE cells Since RPE cells form one aspect of the BRB and are presumed to function in the nutrition of photoreceptor cells and other cells of the neural retina, they must release iron for onward migration to the receptors of those cells. In order to determine whether such release occurred, RPE cells were loaded for 4 hours with 55Fe using 55Fe-Tf,washed free of radiolabel, and then incubated in serum-free medium for various periods of time. Figure 5A shows that under these conditions, about 15% of the accumulated iron was rapidly released during the first hour of incubation. However, no further
PIGMENT EPITHELIAL CELL IRON METABOLISM
0
loo
20
-
40
80
80
100
-
20
r
16
-
120
105
0
0
20
40
60
Bo
100
120
Time (minutes)
B 0
20
0 0
0
J
0
20
40
60
eo
loo
120
Time (minutes) Fig. 4. Transit of internalized "Fe through a small molecular weight pool. A RPE cells were incubated with "'Fe-Tf for various periods of time, washed in glycine-HC1 (pH 3.5),100 mM NaCl at 4°C to remove surface-bound Tf and then lysed by freeze-thawing in PBS. The cytoplasmic extract was separated into a high and a low molecular weight component by centrifugation through a Centricon membrane filter with a molecular weight cut off of 30 kD. The radioactivity associated with the fractions was measured by scintillation counting. (0)Radioactivity associated with material that passes through the membrane. ( 0 ) Radioactivity associated with material that is retained by the membrane. B: The proportion of the radioactivity associated with each fraction a s a function of time.
release was detected. If the medium was changed after the first hour, a further release of 55Feoccurred so that about 40% of the accumulated iron returned to the medium; additional medium chan es, however, did not result in any greater amount of Fe released (Fig. 5B). These data suggest that there is a releasable and a non-releasable intracellular pool of iron and that the former can release more material if the extracellular iron level is lowered. It is, therefore, likely that there is an equilibrium of this pool across the membrane. In addition to changes in medium, the inclusion of serum results in greater release of 55Feby preloaded RPE cells (Fig. 5A). Serum contains iron-binding molecules (including apo-Tf) which may be acting as chelators of released iron, thereby altering the equilibrium across the membrane. When similar incubations of ironloaded cells were carried out in the presence of various concentrations of the specific iron chelator desferrioxamine, again an increase in iron release was observed (Fig. 6). It should be noted that the released iron is not coming from uninternalized surface-adsorbed 55Fe-Tf
Q
o l/ t
a
O Y
0
I
1
2
3
4
5
6
Time (hours) Fig. 5. Release of 55Feby RPE cells. A RPE cells were incubated with V e - T f for 4 hours and were then washed free of unbound Tf at 4°C. The cells were incubated for various periods of time in serum-free ( A ) or serum-containing ( 3 ) medium and radioactivity released into the medium measured by scintillation counting. B: RPE cells were incubated with 55Fe-Tf, washed, and reincubated in serum-free medium with BSA. ( 0 )Released radioactivity without changing the medium. ( A ) Released radioactivity with changes of medium a t the times marked by arrows.
since similar results are obtained when preloaded cells are washed with a low pH salt solution that dissociates any surface bound Tf (data not shown). In addition, no evidence of cell lysis could be observed by staining with trypan blue (data not shown). In order to determine the nature of the recycling of 55Feby RPE cells, uptake and release were measured in the presence of various inhibitory agents (Fig. 7). Energy production inhibitors (2-deoxyglucoseand sodium azide) and acidification inhibitors (monensin and chloroquin) both drastically reduced uptake but had little effect on release while colchicine, which causes the disassembly of microtubules, had no effect on either process.
Release of 55Feto the medium from a transient pool Iron is usually stored intracellularly in association with ferritin, a cytoplasmic protein, having been dissociated from Tf in acidified endosomes (Hanover and Dickson, 1985; Klausner et al., 1983a; Dautry-Varsat et al., 1983: Ciechanover et al., 1983). Since internal-
HUNT AND DAVIS
106
per
30
-
24
-
18-
cent 12
-
6 -
00
-9
-7
-5
-3
log chelator concentration
Fig, 6. Effect of desferrioxamine on release of "Fe by RPE cells. RPE cells were incubated with T e - T f for 4 hours, washed, and incubated for 2 hours in serum-free medium with BSA and the indicated concentrations of defiferrioxamine (hatched bars) or without desferrioxamine (solid bar). Radioactivity in the medium was measured.
Uptake
a
ReIease
grn ab.
30 min release
1 hour release
4 hours 20 hours
release
release
Fig. 8. Dependence of release of 55Feon loading time. RPE cells were incubated 30 minutes (solid bars), 4 hours (bars with small hatched pattern), or 20 hours (large hatched pattern) with "Fee-Tf. They were then washed in 0.2 M glycine-HC1 (pH 3.5) containing 100 mM NaCl and allowed to recycle "Fe into the medium for 0.5,1,4, or 20 hours.
served in the experiment in Fig. 4) represents these carriers and these might also form a releasable iron pool in addition to that complexed to ferritin. As noted above (Fig. 4),the small molecular wei h t (< 30 kD) pool contains a higher proportion of the Fe after short labeling times than after longer periods in which much of the iron has complexed with ferritin. Thus it might be expected that if cells were prelabeled for a period of 30 minutes, a larger proportion of the iron would be available for release than if cells were labeled for 20 hours. Figure 8 shows that this is so. In various experiments, between 30 and 40% of the iron could be released from cells preloaded for just 30 minutes but only about 10% was available for release after a loading time of 20 hours. Nevertheless, the proportion released after 20 hours is greater than the proportion of the label that is in the small molecular weight pool inside the cell, suggesting that ferritin can supply released 55Fe as well.
F,
The nature of the released
iron-containingmaterial Cultured RPE cells were incubated with "Fe-Tf or with "'I-Tf for 4 hours and were then allowed to release label for 2 hours. The released material was sub0 jected to gel filtration on Sephadex G100. As expected, recycled 1251-labeledmaterial eluted in the void volume Fig. 7. Effect of metabolic inhibitors on uptake and release of "Fe from Tf. Cells were incubated with "Fee-Tf in the presence (hatched of the column (Fig. 9C), that is a t the same position a s 1251-Tf(Fig. 9A). The 55Fe-Tfalso migrated in the void bars) or absence (solid bars) of a mixture of 2-deoxyglucose (500 mM) and sodium azide (0.2%) (top) or were preloaded with "Fe from volume (Fig. 9B) but the released "Fe-labeled material "Fe-Tf for 2 hours and allowed to recycle "Fe for 2 hours into serum- eluted partly with the void volume and partly in a secfree medium in the presence (hatched) and absence (solid)of the inhibond rather broad, heterogeneous peak closer to the total itors (bottom).Cell uptake and recycling into the medium of 55Fewas volume of the column, the position at which Dhenol red measured by scintillation counting. is detected (Fig. 9D). Abo& 30% of the released material was found in the void volume and 70% in the included volume of the column. The latter material elutes ized Tf is separated from the cytoplasm by the endoso- more slowly than lysozyme (molecular weight 13 kD) ma1 membrane, there must be other ill-defined iron and extends to the phenol red (molecular weight 354) carriers that participate in transport across the mem- position. When the small molecular weight fraction brane and within the cytoplasm to apo-ferritin. It is was reanalyzed on a Sephadex G25 column which alquite likely that the low molecular weight pool (ob- most excludes insulin (molecular weight 6 kD), much of 10
PIGMENT EPITHELIAL CELL IRON METABOLISM
\
(I)
mr
1%
C
w
c
r
D
t
107
a
b c
d
80
P
60
J
0
0
10
20
30
40
50
Fraction Number
Fig. 10. Sephadex G25 gel filtration of the 55Fe-labeled material. RPE cells were incubated with "Fe-Tf, washed, and allowed to excrete iron-containing compounds into the medium. The material that passed through a 30 kD cut off Centricon filter was analyzed on a Sephadex G25 column. bd, indicates the position of blue dextran; pr, indicates the position of phenol red.
this material also ran with phenol red but three additional peaks ahead of the included fractions were observed (Fig. 10). This indicates multiple components in the small molecular weight fraction. The material that is excluded by this column is also smaller than 10 kD since i t passes through a 10 kD cut off dialysis filter (data not shown). The released material of large molecular weight co-migrates with ferritin When cells were incubated with lz5I-Tf, washed to remove surface-bound radioactivity, and allowed to re-
Fig. 11. Non-denaturing polyacrylamide gel analysis of material recycled by RPE cells. RPE cells were incubated with "'I-Tf or "Fe-Tf for 4 hours, washed in glycine-HC1 (pH 3.5)/100mM NaCI, and allowed to recycle for 2 hours. A lane l, Iz6II-Tfapplied to the cells; lane 2, recycled '"I-apo-Tf; lane 3,lZ"I-labeledmaterial inside the cells. B: lane 1, applied 55Fe-Tf;lane 2, intracellular 55Fe-labeled material; lane 3, recycled j5Fe-labeled material.
lease radiolabeled material, the excreted radioactivity migrated on a non-denaturing polyacrylamide gel at the same position as Tf. 1251-labeledmaterial could also be found within the RPE cells and this too migrated with Tf (Fig. 11,panel A). When a similar experiment was carried out using 55Fe-Tfto label the cells, none of the released material comigrated with Tf (Fig. 11B,
HUNT AND DAVIS
108
Fig. 12. Binding of low molecular weight recycled material to ApoTf. RPE cells were incubated with "Fe-Tf for 4 hours, washed, and allowed to recycle 55Fe into the medium for 2 hours. The labeled low molecular weight material was separated from the high molecular weight material by filtration through a Centricon 30 kD cut off filter and aliquots wcre incubated with 0 (lane l),2 (lane 2), 5 (lane 3),and 10 pg (lane 4) apo-Tf. The contents of the incubation were analyzed directly by non-denaturing PAGE. The gel was dried and autoradiographed after soaking in salicylic acid.
compare lanes 1 and 3 ) but comigrated with the major cytoplasmic "Fe-labeled protein (Fig. 11B, lane 2). The latter (cytoplasmic)protein is probably ferritin since it migrates with authentic spleen ferritin on the same gel. In order t o determine whether the released high molecular weight material was itself ferritin, it was incubated with a polyclonal anti-horse spleen ferritin antibody or a monoclonal anti-human spleen ferritin antibody but in neither case did any precipitation occur; however, both antibodies precipitated the majority of the cytoplasmic ferritin from the same cell cultures. It thus appears that the high molecular weight material that is released from the cells differs in its antigenic properties from normal cytoplasmic ferritin (see Discussion).
Low molecular weight recycled material can donate iron to apo-Tf The low molecular weight recycled 55Fe-labeledmaterial that passes through a 30 kD cut off filter was incubated with various concentrations of apo-Tf. After these incubations, greater than 90% of the radioactivity could no longer pass through the filter (data not shown) and when analyzed by non-denaturing PAGE the 55Fe was shown to comigrate with authentic Tf (Fig. 12).
DISCUSSION RPE cells, together with retinal capillary endothelial cells, control access of blood-borne substances to the neural retina since both are components of the BRB. Among the nutrients in the circulation that bind to receptors on the basolateral surfaces of RPE cells is Tf (Hunt et al., 1989) and iron, either in association with Tf or with some other chelating agent, must cross the BRB in order to be available for the metabolism of cells
that compose the neural retina. It is thus not surprising that cultured RPE cells can release almost half of the iron taken up from radiolabeled diferric Tf after a short incubation period. This finding is in accord with a few other studies of cultured barrier cells in the testis (Sertoli cells) (Wauben-Penris et al., 1988) and the placenta (Be Wo chorion carcinoma cells) (Van der Ende et al., 1987, 1989). Such recycling is also found in macrophages (Saito et al., 1986; Brock et al., 1984; Esparza, and Brock, 1981)but not in other non-barrier cells (e.g., HeLa cells) that have been tested (Van der Ende et al., 1987). Additional data suggest that iron can undergo transcytosis in a variety of capillary endothelial cells such as in the blood-brain barrier (Jefferies et al., 1984) and the blood-liver barrier (Soda and Tavassoli, 1984; Tavassoli et al., 1986). Two models have been proposed for migration of iron from the circulation across barrier cells to cells beyond the barrier. In endothelial cells there is evidence for transcytosis of the iron-Tf complex (Soda and Tavassoli, 1984) although how the diferric Tf is released from its receptor on the ablumenal surface is ill understood since normally diferric Tf exhibits a high affinity for the TfR and only apo-Ti dissociates from the receptor at physiological pH (Klausner et a1., 1983a; Dautry-Varsat et al., 1983). An interesting possibility is the finding that Tf that has undergone endothelial cell transcytosis in the liver has lost some of its sialic acid residues (Tavassoli et al., 1986) which may possibly lower its affinity for the TfR. It is clear, however, that this is not the mechanism by which the majority of the iron is released from RPE cells since recycled iron-labeled Tf could not be detected; indeed, transit of the labeled Tf through a low pH endosomal compartment has occurred which seems to remove both ferric ions from diferric Tf. The ferric ions removed from Tf enter a transient pool of iron that is small in molecular size. This pool in the cytoplasm may be able to equilibrate across the plasma membrane as well as supplying iron to ferritin although there is no evidence that the small (non-ferritin) cytoplasmic iron chelator is the same as that which appears outside the cell. The excreted "Fe most likely comes from the transient low molecular weight pool because the externalized "Fe is itself of low molecular weight and the fraction of the total cell label that is externalized is related t o the fraction of the internal labeled material that is low molecular weight. The precise nature of the low molecular weight material is not known but the released material is not free ferric ions because their concentration is much higher than the maximum solubility of free ferric iron at neutral pH (10- l7 MI. Moreover, free ferric ions do not migrate in such a heterogeneous manner on gel filtration. Thus the iron must be chelated by a small molecule. The affinity of this molecule for iron is less than apo-Tf since the latter binds the released 55Fe almost quantitatively when added to it. The precise nature of the externalized iron chelate is presently under investigation. A puzzling finding reported here is the release of a high molecular weight iron-containing protein that migrates on gels in a manner similar to ferritin. Similar proteins hae been seen in material recycled by Be Wo cells (Van der Ende et al., 1987), Sertoli cells (Wauben-
PIGMENT EPITHELIAL CELL IRON METABOLISM
Penris et al., 1988), and macrophages (Saito et al., 1986; Brock e t al., 1984; Esparza and Brock, 1981) and have, in some cases, been ascribed to a small amount of cell lysis. This is certainly possible since, even after short loading periods, the proportion of total cell 55Fein ferritin i s large. Nevertheless, staining of the cells at the end of the recycling period with trypan blue showed no evidence whatsoever of cell death. This apparent lack of lysis during recycling has also been observed by others who reached the conclusion that lysis could not result in the observed level of secreted iron-binding proteins (Brock et al., 1984). Since some 40% of the “5Fe accumulated by cells in a period of 4 hours is recycled to the environment and about a third of this is the high molecular weight component, this would indicate enough cell death to be readily observed. An alternative explanation is that the high molecular weight material is secreted by the cells without cell lysis as part of the normal iron recycling procedure. It is known that ferritin can be secreted into the serum (Munro and Linder, 1978) and that cardiac cells have ferritin mRNA associated with membrane-associated ribosomes (Campbell et al., 1989). However, serum ferritin is a glycoprotein which binds to concanavalin A (Santambrogio et al., 1987; Cragg e t al., 1981; Wonvood et al., 1979) and we can find no evidence of glycosylation of the high molecular weight ferritin-like material from RPE cells. One piece of evidence supports the contention that this material is not just cytoplasmic ferritin from lyzed cells: It has been found that cytoplasmic ferritin can be immunoprecipitated by a monoclonal and a polyclonal anti-ferritin antibody but the released material cannot. This is not the result of a n inhibitor of antibody binding in the released material since when radiolabeled cytoplasmic ferritin is mixed with recycled material, 55Feis immunoprecipitated in proportion to the cytoplasmic component. It is quite possible that the phenomena described are part of the mechanism of iron transport to the photoreceptor cells since the eye, particularly the aqueous (Tripathi et al., 1990; Yu and Okamura, 1988) and vitreous humor (Hawkins, 1986), contains substantial amounts of apo-Tf. In addition, recent evidence suggests that cells of the neural retina can synthesize W themselves since Tf mRNA has been detected within them (A. Swaroop, personal communication). Current investigations in our laboratory are attempting to determine precisely which cells in the neural retina make apo-Tf. The apo-Tf made within the retina could bind iron released from the apical surfaces of RPE cells and diffuse to the photoreceptor cell inner segments which we have shown to bear TfRs (R.C. Hunt and A.A. Davis, unpublished observations). This mechanism together with the close apposition of photoreceptors to the RPE cells would overcome the problems of diffusion of iron-bearing Tf through the vitreous humor and the layers of the neural retina to the highly metabolic photoreceptor cells. The nature of the high and low molecular weight iron-containing material secreted by human RPE cells remains to be determined; however, our findings of substantial iron release by RPE cells and the binding of this material to apo-Tf suggest a model to explain the transport of iron from the bloodstream across the BRB
109
to iron-requiring cells of the retina. Additional support is lent to this idea by the observations that neural retina cells can make apo-Tf (Hunt and Davis, unpublished observations; Swaroop, personal communication) and that photoreceptors can bind i t (Hunt and Davis, unpublished observations).
ACKNOWLEDGMENTS We thank the South Carolina Lions’ Eye Bank for supplying human eye material for cell culture. This work was supported by grant EY06164 from the National Institutes of Health. LITERATURE CITED Barnes, D., and Sato, G. (1980) Serum-free cell culture: A unifying approach. Cell, 22r649-655. Brock, J.H., Esparza, I., and Logie, A.C. (1984) The nature of iron released by resident or stimulated mouse peritoneal macrophages. Biochim. Biophys. A d a , 797t105-111. Campbell, C.H., Ismail, R., and Linder, M.C. (1989)Ferritin mRNA is found on bound as well as on free polyribosomes in rat heart. Arch. Biochem. Biophys., 273t89-98. Ciechanover, A., Schwartz, A.L., Dautry-Varsat, A., and Lodish, H.F. (1983) Kinetics of internalization and recycling of transferrin and the transferrin receptor in a human heptoma cell line. J. Biol. Chem., 258t968 1-9689. Clark, V.M. (1986) The cell biology of the retinal pigment epithelium. In: The Retina: A Model for Cell Biology Studies. Part 11. R. Adler, and D. Farber, eds. Academic Press, Orlando, Florida, pp. 129-169. Cragg, S.J., Wagstaff, M., and Worwood, M. (1981) Detection of glycosylated subunit in human serum ferritin. Biochem. J., 199t565-571. Cunha-Vaz, J. (1979) The blood-ocular barriers. Surv. Ophthalmol., 23:279-296. Dautry-Varsat, A., Ciechanover, A., and Lodish, H.F. (1983) pH and the recycling of transferrin during receptor-mediated endocytosis. Proc. Natl. Acad. Sci. U.S.A., 80t2258-2262. Esparza, I., and Brock, J.H. (1981) Release of iron by resident and stimulated mouse peritoneal macrophages following ingestion and degradation of transferrin-anti-transferrin immune complexes. Br. J. Haematol., 49t603-614. Fuller, S.D., and Simons, K. (1986) Transferrin receptor polarity and recycling accuracy in “tight” and “leaky” strains of Madin-Darby canine kidney cells. J . Cell Biol., 103t1767-1779. Hanover, J.A., and Dickson, R.B. (1985) Transferrin: Receptor-mediated endocytosis and iron delivery. In: Endocytosis. I. Pastan, and M.C. Willingham, eds. Plenum Press, New York and London, pp. 131-161. Hawkins, K.N. (1986) Contributions of plasma proteins to the vitreous of the rat. Curr. Eye Res., 5:655-663. Hunt, R.C., Ruffin, R., and Yang, Y-S. (1984)Alterations in the transferrin receptor of human erythroleukemic cells after induction of hemoglobin synthesis. J. Biol. Chem.,259t9944-9952. Hunt, R.C., Dewey, A,, and Davis, A.A. (1989)Transferrin receptors of the retinal pigment epithelium are associated with the cytoskeleton. J. Cell Sci., 92:655-666. Jefferies, W.A., Brandon, M.R., Hunt, S.V., Williams, A.F., Gatter, K.C., and Mason, D.Y. (1984) Transferrin receptor on endothelium of brain capillaries. Nature, 312t162-163. Klausner, R.D., Ashwell, G., van Renswoude, J., Harford, J.B., and Bridges, K.R. (1983a) Binding of apotransferrin to K562 cells: Explanation of the transferrin cycle. Proc. Natl. Acad. Sci. U.S.A., 80:2263-2266. Klausner, R.D., van Renswoude,J.,Ashwell, G., Kempf, C., Schechter, A.N., Dean, A., and Bridges, K. (198313)Receptor-mediated endocytosis of transferrin by K562 cells. J. Biol. Chem., 258:47154724. Laemmli, U.K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature, 227t680-685. Lakhanpal, V., Schacket, S., and Rouben, J. (1984) Desferrioxamine (Desferabinduced toxic retinal pigmentary degeneration and presumed optic neuropathy. Ophthalmology, 91t443-451. Munro, H.N., and Linder, M.C. (1978)Ferritin structure, biosynthesis and role in iron metabohsm. Physiol. Rev., 58:317396. Pedersen, 0. (1979)An electron microscopic study of the permeability of intraocular blood vessels using lanthanum as a tracer in vivo. Exp. Eye Res., 21t61-69.
110
HUNT AND DAVIS
Peyman, G.A., and Bok, D. (1972).Peroxidase diffusion in the normal and laser-coagulated primate retina. Invest. Ophthalmol., 21:35-
45. Saito, K., Nishisato, T., Grasso, J.A., and Aisen, P. (1986) Interaction of transferrin with iron-loaded rat peritoneal macrophages. Br. J . Haematol., 62:275-286. Santambrogio, P., Cozzi, A,, Levi, S., and Arioso, P. (1987) Human serum ferritin G-peptide is recognized by anti-L ferritin subunit antibodies and concanavalin A. Br. J. Haematol., 65.235-237. Shichi, H. (1969) Microsomal electron transport system of bovine retinal pigment epithelium. Exp. Eye Res., 8r60-68. Smith, R.S., and Rudt, L.A. (1975) Ocular vascular and epithelial barriers to niicroperoxidase. Invest. Ophthalmol., 14:55&560. Soda, R., and Tavassoli, M. (1984) Transendothelial transport (transcytosis) of iron-transferrin complex in the bone marrow. J . 1Jltrastruct. Res., 88:18-24. Tavassoli, M.: Kishimoto, T., Soda, R., Kataoka, M., and Harjes, K. (1986) Liver endothelium mediates the uptake of iron-transferrin complex by hepatocytes. Exp. Cell Res., 165:369379. Tripathi, R.C., Millard, C.B., Tripathi, B.J., Chailertborisuth, N.S.,
Neeley, K.A., and Ernest, J.T. (1990)Aqueous humor ofcat contains fibroblast growth factor and transferrin similar to those in man. Exp. Eye Res., 50:10%112. Van der Ende, A., Du Maine, A,, Simmons, C.F., Schwartz, A.L., and Strous, G.J. (1987) Iron metabolism in Be Wo chorion carcinoma cells: Transferrin-mediated uptake and release of iron. J. Biol. Chem., 262t8910-8916. Van der Ende, A., Du Maine, A,, Schwartz, A.L., and Strous, G.J. (1989)Effect of ATP depletion and temperature on the transferrinmediated uptake and release of iron by Be Wo choriocarcinomacells. Biochem. J.,259:685-692. Wauben-Penris, P.J.J., Veldscholte, J.,Van der Ende, A., and van der Donk, H.A. (1988)The release of iron by Sertoli cells in culture, Biol. Reprod., 38: 1105-1 113. Worwood, M., Cragg, S.J.,Wagstaff, M., and Jacobs, A . (1979)Binding of serum ferritin to concanavalin A. Clin. Sci., 56.83-87. Yu, T-C., and Okamura, R. (1988)Quantitative study of characteristic aqueous humor transferrin, serum- transferrin and desialized transferrin in aqueous humor. Jpn. J. Ophthalmol., 32:26%274.
Release of Iron by Human Retinal Pigment Epithelial Cells RICHARD C. H U N T * AND ALBERTA A. DAVIS Department of Microbiology, Univeriity of South Carolina Medicdl School, Columbia, South Carolina 29208 Retinal pigment epithelial cells, which form one aspect of the blood-retinal barrier, take up iron in association with transferrin by a typical receptor-mediated mechanism (Hunt et al., 1989. J. Cell Sci. 92:655-666). This iron is dissociated from transferrin in a low pH environment and uptake i s sensitive to agents that inhibit endosomal acidification. The dissociated iron enters the cytoplasm as a low molecular weight (< 10 kD) component and subsequently binds to ferritin. No evidence for recycling of iron in association with transferrin was found. Nevertheless, much of the iron that is taken up is recycled to the extracellular medium, primarily from the low molecular weight pool. This release of iron is not sensitive to inhibitors of energy production or of vesicular acidification but is increased u p to a maximum of about 40% of the total 55Feincorporated when cells are incubated with serum or the medium is changed. When a short loading time for ""Fe from 55Fe-transferrin is used (i.e., when the low molecular weight pool i s proportionately larger), a much larger fraction of the cell-associated radiolabel is released than when longer loading times are used. The data suggest that a releasable intracellular iron pool is in equilibrium with the externalized material. The released iron may be separated into a high and a low molecular weight component. The former i s similar on polyacrylamide gel electrophoresis to ferritin although it cannot be immune precipitated by anti-ferritin antibodies. The low molecular weight 55Fewhich i s heterogeneous in nature can be bound by external apo-transferrin and may represent a form that can be taken up by cells beyond the blood-retinal barrier. o 1992 Wiiev-Liss, Inc.
Free diffusion of blood-borne substances into the neu- ceptor cells since although they do not divide, they ral retina is obstructed by barrier layers of cells that carry out extensive membrane production in order to give the eye its immune privileged properties. The compensate for the shedding of the disk membranes blood-retinal barrier (BRB) comprises the tight junc- that contain the photoreceptor pigment. Thus, it is not tions of the endothelial cells of the retinal capillaries surprising that lack of iron has been suggested t o lead and of the retinal pigment epithelium (Pedersen, 1979; t o retinal dysfunction (Lakhanpal et al., 1984). Peyman and Bok, 1972; Smith and Rudt, 1975) and any The mechanism by which iron is taken up at the nutrient, hormone, or growth factor that must enter the basolateral surface of RPE cells, transported to their neural retina must first cross these cells. Thus selectiv- apical surface, and released for onward migration to the ity of access to, for example, the photoreceptor cells is neural retina is unknown though transferrin receptors determined by whether BRB cells can bind and trans- (TfRs)of the type found on almost all other cells in the port that substance. body are located on the basolateral surfaces of RPE The retinal pigment epithelium (RPE)is a monolayer (Hunt et al., 1989) and other epithelial cells (Fuller and of polarized cells containing melanin and lipofuscin Simons, 1986). After binding and internalization of Tf (Clark, 1986). Their basolateral surface is in contact a t the basolateral surfaces of RPE cells, one of two with the Bruch's membrane and the circulation via the mechanisms for onward transport is most likely. In one fenestrated capillaries of the choroid while their apical model (Fig. lA), Tf is internalized in vesicles that misurfaces embrace the outer segments of the photorecep- grate to the apical surface where it is released and tors. Intercellular diffusion is blocked by tight junc- binds to TfRs on photoreceptor and other cells. In a tional complexes that girdle the cell (Cunha-Vaz, second more complicated model (Fig. lB), Tf loses its 1979). iron in the low pH endosomes of RPE cells and apo-Tf is Iron, carried in the circulation in association with the recycled to the circulation. The iron enters the cytocarrier protein, transferrin (TO,is an obligatory growth factor for all cells (Barnes and Sato, 19801, being involved in a variety of processes that require iron-containing enzymes. Among the latter are fatty acid desaturases (Shichi, 1969) that are important in membrane Received J u l y 30, 1991; accepted February 10, 1992. biogenesis. These enzymes are important to photore- *To whom reprint requests/correspondence should be addressed. 0 1992 WI1,EY-LISS, INC.
PIGMENT EPITHELIAL CELL IRON METABOLISM
apical
e
troloacetic acid and sodium bicarbonate (Klausner et al., 1983b). Free Na’”1 or “FeC1, were removed by Sephadex G25 gel filtration.
e
Q
A
b
B
Q t
Tfe
basolateral
103
apoTf
0 0
Fig. 1. Transcytosis of iron by RPE cells. Model A (left):Diferric Tf (filled circles) binds to a TfR on the hasolateral surface of a n RPE cell, The receptormf complex is internalized and transported to the apical surface where the Tf dissociates for onward migration to the cells of the neural retina. Model B (right):Diferric Tf hinds to a TfR on the hasolateral surface of the cell and is internalized into a low pH endosome where iron (slashed circles) dissociates from the Tf. The apo-Tf (open circles) recycles to the basolateral surface of the cells where it is released from the TfR. The released iron is transported across the membrane of the endosome, probably hound to a hypothetical carrier and enters the cytoplasm. Here it dissociates from the carrier and binds to cytoplasmic apo-ferritin (squares). The cytoplasmic iron pool, either in association with ferritin or the hypothetical carrier molecule, is the source of iron that is secreted from the apical surface of the cell and which binds to apo-Tf in the interphotoreceptor matrix. This differric Tf (filled circles) then hinds to TfRs on cells of the neural retina. The cytoplasmic iron that is not complexed to ferritin and that is secreted into the interphotoreceptor matrix may be bound by the same or different carrier molecules.
plasm where it is complexed to ferritin or some other cytoplasmic chelating agent and this cytoplasmic pool is the source of iron that is secreted at the apical surface for the use of cells beyond the BRB. In this paper, evidence for the latter model is presented.
MATERIALS AND METHODS Culture of RPE cells RPE cells from human donor eyes were grown in Ham’s F10 medium as previously described (Hunt et al., 1989). Preparation of radiolabeled Tf Diferric Tf (ICN Biomedical) was dissolved in phosphate buffered saline (PBS) and iodinated using chloramine T and Na1251(Amersham) as described (Hunt et al., 1984). Apo-Tf (Boehringer-Mannheim) was labeled with 55FeC1, (Amersham) in the presence of ni-
Abbreviations BRB. blood-retinal barrier: BSA. bovine serum albumin; PAGE, polvacrylamide gel electrophoresis; PBS, phosphate buffered saline; RPE, retinal pigment epithelial; SDS, sodium dodecyl sulfate; Tf, transferrin; lYR, transferrin receptor
Non-denaturing polyacrylamide gel electrophoresis 55 Fe-labeled material was analyzed by polyacrylamide gel electrophoresis (PAGE) using the buffer system of Laemmli (Laemmli, 1970) but omitting sodium dodecyl sulfate (SDS) from the electrode buffers and the gel. Labeled protein was applied directly to the gel without SDS or dithiothreitol. After electrophoresis, the gel was fixed in methanol/water (1:l) for 30 minutes and incubated in salicylic acid solution for 20 minutes (Hunt et al., 1984). The gel was then dried and autoradiographed. Incubation of RPE cells The conditions for individual experiments are described in the legends to the Figures. In all cases, binding and uptake of radiolabeled Tf were carried out in Dulbecco’s modified Eagle’s (DME) medium (Gibco) supplemented with penicillin, streptomycin, and bovine serum albumin (BSA) (1 mg/ml). Non-specific binding and uptake were measured in the presence of 100 fold excess unlabeled Tf. The non-specific binding was less than 10% of the specific binding. Removal of surface-boundTf After incubation of cells with radiolabeled Tf, the cells were washed three times at 4°Cwith PBS and then incubated a t 4°C with 0.2 M glycine HC1 (pH 3.5) containing 100 mM NaCl for 5 minutes. The cells were washed again in PBS. Release of cytoplasmic components 0.5 ml PBS was added to a monolayer of washed cells which were then frozen at -20°C. After thawing, the cells were scraped from the bottle with a rubber policeman and centrifuged at 15,000 rpm (16,OOOg) on a Beckman Microfuge for 5 minutes to yield a crude soluble fraction. The small molecular weight molecules were separated from the remainder of the fraction by centrifugation through a Centricon 30 or 10 kD dialysis filter (Amicon). Separation of low and high molecular weight material excreted from RPE cells Medium into which cells had recycled radiolabeled material was centrifuged through Centricon 30 kD or 10 kD cut off filters (Amicon). RESULTS Uptake of Tf and iron by RPE cells The protein moiety of Tf was labeled with 1251and incubated with cultured human RPE cells for various periods of time at 37°C. The cell-associated Tf was determined to be either surface or internal by removing the former in a low pH wash. Non-specific binding was assessed in parallel incubations in the presence of excess unlabeled Tf and was found to be less than 10% of the specific uptake. All non-specific binding appeared to be cell surface (released by low pH) with none internalized. Figure 2A shows that both surface and internal
HUNT AND DAVIS
104 300-
*O0
r
/
700
-
240
=
'
a,
180
-
600
500
0
6
0
6
0
20
40
60
60
120
180
240
100
80
Time (minutes) 0
240
180
Fig. 3. Uptake of 55Fe from 55Fe-Tf by RPE cells. RPE cells were incubated with 55Fe-Tfor "'I-Tf for various periods of time and were treated in a manner similar to those in Figure 2. ( 0 ) 1251-Tf. (A) 55Fe-Tf.( 0 )55Fe-Tfin the presence of 10 pM monensin.
1 I
120
0
t 0
partment was shown by the use of monensin which totally abolished iron accumulation by RPE cells (Fig. 3, solid symbols).
20
40
60
80
100
Time (minutes) Fig. 2. Uptake of I2'II-Tf by RPE cells. A: RPE cells in sub-confluent monolayer culture were incubated with 1251-Tfin serum-free medium supplemented with BSA (1 mg/ml) for various periods of time. After washing the cells at 4"C, they were incubated with low pH buffered saline (0.2M glycine HCI, pH 3.5 containing lO0nM NaCl) to remove surface-associated Tf. The cells were then dissolved in 0.1 M NaOH. Radioactivity associated with the acid wash and the cells was measured by T counting. ( 0 ) Internal "'I. ( 0 ) Surface lZ51.B: The effect of monensin (10KM)on the uptake of '""I-Tfwas measured in a similar experiment. In all cases total binding is corrected for non-specific binding by subtracting the binding in the presence of 100 fold excess unlabeled Tf. (0)Internal Iz5I.( 0 ) Surface lZ5I.
Tf were accumulated by the cells and that accumulation reached a plateau level by about an hour. The saturation of uptake at the plateau level is similar to observations with other cells in which apo-Tf is recycled to the cells' environment (Hanover and Dickson, 1985). At the plateau level there is equilibrium between uptake and recycling. To determine whether the apparent recycling of apo-Tf by RPE cells occurs via a low pH endosomal compartment, cells were incubated with ?-Tf in the presence of monensin which inhibits endosomal acidification (Fig. 2B). In this case, surface binding and cellular accumulation are both reduced by about 60%,resulting probably from the inhibition of TfR recycling in the absence of acidification. In contrast to lz5I-Tf, the cellular accumulation of "Fe-Tf did not reach a plateau (at least up to 4 hours) (Fig. 3) although, as expected, surface binding did exhibit a plateau value (data not shown). Thus, iron must be removed from the Tf for intracellular accumulation. That this was due to passage through an acidic com-
Transit of 5"Fethrough a low molecular weight pool In order to assess the nature of the molecules to which 55Feis bound after endocytosis of "Fe-labeled Tf, cells were incubated for various periods of time (1 minute to 2 hours) with 55Fe-Tf, washed in low pH solution to remove any surface bound Tf, and then frozen and thawed to release cytoplasmic components. The cytoplasmic fraction was then passed through a membrane filter with a molecular weight cut off of 30 kD to separate small and large iron-binding molecules, At early time points, the majority of the Fe is associated with a molecule that passes through the filter. The pool of this small molecular weight cytoplasmic component is rapidly saturated with "Fe (Fig. 4A). The proportion of the label that is in this fraction falls rapidly as iron is accumulated in a > 30 kD cytoplasmic component that does not pass through the filter (Fig. 4B). This material comigrates with ferritin on non-denaturing PAGE (data not shown). The "Fe-labeled small molecular weight material associated with the cells is not simply the adherence of an impurity in the 55Fe-Tfpreparation since i) it does not occur a t 4"C, ii) it is abolished by excess unlabeled Tf, and iii) it is not removed by a low pH incubation. Release of iron to the medium by RPE cells Since RPE cells form one aspect of the BRB and are presumed to function in the nutrition of photoreceptor cells and other cells of the neural retina, they must release iron for onward migration to the receptors of those cells. In order to determine whether such release occurred, RPE cells were loaded for 4 hours with 55Fe using 55Fe-Tf,washed free of radiolabel, and then incubated in serum-free medium for various periods of time. Figure 5A shows that under these conditions, about 15% of the accumulated iron was rapidly released during the first hour of incubation. However, no further
PIGMENT EPITHELIAL CELL IRON METABOLISM
0
loo
20
-
40
80
80
100
-
20
r
16
-
120
105
0
0
20
40
60
Bo
100
120
Time (minutes)
B 0
20
0 0
0
J
0
20
40
60
eo
loo
120
Time (minutes) Fig. 4. Transit of internalized "Fe through a small molecular weight pool. A RPE cells were incubated with "'Fe-Tf for various periods of time, washed in glycine-HC1 (pH 3.5),100 mM NaCl at 4°C to remove surface-bound Tf and then lysed by freeze-thawing in PBS. The cytoplasmic extract was separated into a high and a low molecular weight component by centrifugation through a Centricon membrane filter with a molecular weight cut off of 30 kD. The radioactivity associated with the fractions was measured by scintillation counting. (0)Radioactivity associated with material that passes through the membrane. ( 0 ) Radioactivity associated with material that is retained by the membrane. B: The proportion of the radioactivity associated with each fraction a s a function of time.
release was detected. If the medium was changed after the first hour, a further release of 55Feoccurred so that about 40% of the accumulated iron returned to the medium; additional medium chan es, however, did not result in any greater amount of Fe released (Fig. 5B). These data suggest that there is a releasable and a non-releasable intracellular pool of iron and that the former can release more material if the extracellular iron level is lowered. It is, therefore, likely that there is an equilibrium of this pool across the membrane. In addition to changes in medium, the inclusion of serum results in greater release of 55Feby preloaded RPE cells (Fig. 5A). Serum contains iron-binding molecules (including apo-Tf) which may be acting as chelators of released iron, thereby altering the equilibrium across the membrane. When similar incubations of ironloaded cells were carried out in the presence of various concentrations of the specific iron chelator desferrioxamine, again an increase in iron release was observed (Fig. 6). It should be noted that the released iron is not coming from uninternalized surface-adsorbed 55Fe-Tf
Q
o l/ t
a
O Y
0
I
1
2
3
4
5
6
Time (hours) Fig. 5. Release of 55Feby RPE cells. A RPE cells were incubated with V e - T f for 4 hours and were then washed free of unbound Tf at 4°C. The cells were incubated for various periods of time in serum-free ( A ) or serum-containing ( 3 ) medium and radioactivity released into the medium measured by scintillation counting. B: RPE cells were incubated with 55Fe-Tf, washed, and reincubated in serum-free medium with BSA. ( 0 )Released radioactivity without changing the medium. ( A ) Released radioactivity with changes of medium a t the times marked by arrows.
since similar results are obtained when preloaded cells are washed with a low pH salt solution that dissociates any surface bound Tf (data not shown). In addition, no evidence of cell lysis could be observed by staining with trypan blue (data not shown). In order to determine the nature of the recycling of 55Feby RPE cells, uptake and release were measured in the presence of various inhibitory agents (Fig. 7). Energy production inhibitors (2-deoxyglucoseand sodium azide) and acidification inhibitors (monensin and chloroquin) both drastically reduced uptake but had little effect on release while colchicine, which causes the disassembly of microtubules, had no effect on either process.
Release of 55Feto the medium from a transient pool Iron is usually stored intracellularly in association with ferritin, a cytoplasmic protein, having been dissociated from Tf in acidified endosomes (Hanover and Dickson, 1985; Klausner et al., 1983a; Dautry-Varsat et al., 1983: Ciechanover et al., 1983). Since internal-
HUNT AND DAVIS
106
per
30
-
24
-
18-
cent 12
-
6 -
00
-9
-7
-5
-3
log chelator concentration
Fig, 6. Effect of desferrioxamine on release of "Fe by RPE cells. RPE cells were incubated with T e - T f for 4 hours, washed, and incubated for 2 hours in serum-free medium with BSA and the indicated concentrations of defiferrioxamine (hatched bars) or without desferrioxamine (solid bar). Radioactivity in the medium was measured.
Uptake
a
ReIease
grn ab.
30 min release
1 hour release
4 hours 20 hours
release
release
Fig. 8. Dependence of release of 55Feon loading time. RPE cells were incubated 30 minutes (solid bars), 4 hours (bars with small hatched pattern), or 20 hours (large hatched pattern) with "Fee-Tf. They were then washed in 0.2 M glycine-HC1 (pH 3.5) containing 100 mM NaCl and allowed to recycle "Fe into the medium for 0.5,1,4, or 20 hours.
served in the experiment in Fig. 4) represents these carriers and these might also form a releasable iron pool in addition to that complexed to ferritin. As noted above (Fig. 4),the small molecular wei h t (< 30 kD) pool contains a higher proportion of the Fe after short labeling times than after longer periods in which much of the iron has complexed with ferritin. Thus it might be expected that if cells were prelabeled for a period of 30 minutes, a larger proportion of the iron would be available for release than if cells were labeled for 20 hours. Figure 8 shows that this is so. In various experiments, between 30 and 40% of the iron could be released from cells preloaded for just 30 minutes but only about 10% was available for release after a loading time of 20 hours. Nevertheless, the proportion released after 20 hours is greater than the proportion of the label that is in the small molecular weight pool inside the cell, suggesting that ferritin can supply released 55Fe as well.
F,
The nature of the released
iron-containingmaterial Cultured RPE cells were incubated with "Fe-Tf or with "'I-Tf for 4 hours and were then allowed to release label for 2 hours. The released material was sub0 jected to gel filtration on Sephadex G100. As expected, recycled 1251-labeledmaterial eluted in the void volume Fig. 7. Effect of metabolic inhibitors on uptake and release of "Fe from Tf. Cells were incubated with "Fee-Tf in the presence (hatched of the column (Fig. 9C), that is a t the same position a s 1251-Tf(Fig. 9A). The 55Fe-Tfalso migrated in the void bars) or absence (solid bars) of a mixture of 2-deoxyglucose (500 mM) and sodium azide (0.2%) (top) or were preloaded with "Fe from volume (Fig. 9B) but the released "Fe-labeled material "Fe-Tf for 2 hours and allowed to recycle "Fe for 2 hours into serum- eluted partly with the void volume and partly in a secfree medium in the presence (hatched) and absence (solid)of the inhibond rather broad, heterogeneous peak closer to the total itors (bottom).Cell uptake and recycling into the medium of 55Fewas volume of the column, the position at which Dhenol red measured by scintillation counting. is detected (Fig. 9D). Abo& 30% of the released material was found in the void volume and 70% in the included volume of the column. The latter material elutes ized Tf is separated from the cytoplasm by the endoso- more slowly than lysozyme (molecular weight 13 kD) ma1 membrane, there must be other ill-defined iron and extends to the phenol red (molecular weight 354) carriers that participate in transport across the mem- position. When the small molecular weight fraction brane and within the cytoplasm to apo-ferritin. It is was reanalyzed on a Sephadex G25 column which alquite likely that the low molecular weight pool (ob- most excludes insulin (molecular weight 6 kD), much of 10
PIGMENT EPITHELIAL CELL IRON METABOLISM
\
(I)
mr
1%
C
w
c
r
D
t
107
a
b c
d
80
P
60
J
0
0
10
20
30
40
50
Fraction Number
Fig. 10. Sephadex G25 gel filtration of the 55Fe-labeled material. RPE cells were incubated with "Fe-Tf, washed, and allowed to excrete iron-containing compounds into the medium. The material that passed through a 30 kD cut off Centricon filter was analyzed on a Sephadex G25 column. bd, indicates the position of blue dextran; pr, indicates the position of phenol red.
this material also ran with phenol red but three additional peaks ahead of the included fractions were observed (Fig. 10). This indicates multiple components in the small molecular weight fraction. The material that is excluded by this column is also smaller than 10 kD since i t passes through a 10 kD cut off dialysis filter (data not shown). The released material of large molecular weight co-migrates with ferritin When cells were incubated with lz5I-Tf, washed to remove surface-bound radioactivity, and allowed to re-
Fig. 11. Non-denaturing polyacrylamide gel analysis of material recycled by RPE cells. RPE cells were incubated with "'I-Tf or "Fe-Tf for 4 hours, washed in glycine-HC1 (pH 3.5)/100mM NaCI, and allowed to recycle for 2 hours. A lane l, Iz6II-Tfapplied to the cells; lane 2, recycled '"I-apo-Tf; lane 3,lZ"I-labeledmaterial inside the cells. B: lane 1, applied 55Fe-Tf;lane 2, intracellular 55Fe-labeled material; lane 3, recycled j5Fe-labeled material.
lease radiolabeled material, the excreted radioactivity migrated on a non-denaturing polyacrylamide gel at the same position as Tf. 1251-labeledmaterial could also be found within the RPE cells and this too migrated with Tf (Fig. 11,panel A). When a similar experiment was carried out using 55Fe-Tfto label the cells, none of the released material comigrated with Tf (Fig. 11B,
HUNT AND DAVIS
108
Fig. 12. Binding of low molecular weight recycled material to ApoTf. RPE cells were incubated with "Fe-Tf for 4 hours, washed, and allowed to recycle 55Fe into the medium for 2 hours. The labeled low molecular weight material was separated from the high molecular weight material by filtration through a Centricon 30 kD cut off filter and aliquots wcre incubated with 0 (lane l),2 (lane 2), 5 (lane 3),and 10 pg (lane 4) apo-Tf. The contents of the incubation were analyzed directly by non-denaturing PAGE. The gel was dried and autoradiographed after soaking in salicylic acid.
compare lanes 1 and 3 ) but comigrated with the major cytoplasmic "Fe-labeled protein (Fig. 11B, lane 2). The latter (cytoplasmic)protein is probably ferritin since it migrates with authentic spleen ferritin on the same gel. In order t o determine whether the released high molecular weight material was itself ferritin, it was incubated with a polyclonal anti-horse spleen ferritin antibody or a monoclonal anti-human spleen ferritin antibody but in neither case did any precipitation occur; however, both antibodies precipitated the majority of the cytoplasmic ferritin from the same cell cultures. It thus appears that the high molecular weight material that is released from the cells differs in its antigenic properties from normal cytoplasmic ferritin (see Discussion).
Low molecular weight recycled material can donate iron to apo-Tf The low molecular weight recycled 55Fe-labeledmaterial that passes through a 30 kD cut off filter was incubated with various concentrations of apo-Tf. After these incubations, greater than 90% of the radioactivity could no longer pass through the filter (data not shown) and when analyzed by non-denaturing PAGE the 55Fe was shown to comigrate with authentic Tf (Fig. 12).
DISCUSSION RPE cells, together with retinal capillary endothelial cells, control access of blood-borne substances to the neural retina since both are components of the BRB. Among the nutrients in the circulation that bind to receptors on the basolateral surfaces of RPE cells is Tf (Hunt et al., 1989) and iron, either in association with Tf or with some other chelating agent, must cross the BRB in order to be available for the metabolism of cells
that compose the neural retina. It is thus not surprising that cultured RPE cells can release almost half of the iron taken up from radiolabeled diferric Tf after a short incubation period. This finding is in accord with a few other studies of cultured barrier cells in the testis (Sertoli cells) (Wauben-Penris et al., 1988) and the placenta (Be Wo chorion carcinoma cells) (Van der Ende et al., 1987, 1989). Such recycling is also found in macrophages (Saito et al., 1986; Brock et al., 1984; Esparza, and Brock, 1981)but not in other non-barrier cells (e.g., HeLa cells) that have been tested (Van der Ende et al., 1987). Additional data suggest that iron can undergo transcytosis in a variety of capillary endothelial cells such as in the blood-brain barrier (Jefferies et al., 1984) and the blood-liver barrier (Soda and Tavassoli, 1984; Tavassoli et al., 1986). Two models have been proposed for migration of iron from the circulation across barrier cells to cells beyond the barrier. In endothelial cells there is evidence for transcytosis of the iron-Tf complex (Soda and Tavassoli, 1984) although how the diferric Tf is released from its receptor on the ablumenal surface is ill understood since normally diferric Tf exhibits a high affinity for the TfR and only apo-Ti dissociates from the receptor at physiological pH (Klausner et a1., 1983a; Dautry-Varsat et al., 1983). An interesting possibility is the finding that Tf that has undergone endothelial cell transcytosis in the liver has lost some of its sialic acid residues (Tavassoli et al., 1986) which may possibly lower its affinity for the TfR. It is clear, however, that this is not the mechanism by which the majority of the iron is released from RPE cells since recycled iron-labeled Tf could not be detected; indeed, transit of the labeled Tf through a low pH endosomal compartment has occurred which seems to remove both ferric ions from diferric Tf. The ferric ions removed from Tf enter a transient pool of iron that is small in molecular size. This pool in the cytoplasm may be able to equilibrate across the plasma membrane as well as supplying iron to ferritin although there is no evidence that the small (non-ferritin) cytoplasmic iron chelator is the same as that which appears outside the cell. The excreted "Fe most likely comes from the transient low molecular weight pool because the externalized "Fe is itself of low molecular weight and the fraction of the total cell label that is externalized is related t o the fraction of the internal labeled material that is low molecular weight. The precise nature of the low molecular weight material is not known but the released material is not free ferric ions because their concentration is much higher than the maximum solubility of free ferric iron at neutral pH (10- l7 MI. Moreover, free ferric ions do not migrate in such a heterogeneous manner on gel filtration. Thus the iron must be chelated by a small molecule. The affinity of this molecule for iron is less than apo-Tf since the latter binds the released 55Fe almost quantitatively when added to it. The precise nature of the externalized iron chelate is presently under investigation. A puzzling finding reported here is the release of a high molecular weight iron-containing protein that migrates on gels in a manner similar to ferritin. Similar proteins hae been seen in material recycled by Be Wo cells (Van der Ende et al., 1987), Sertoli cells (Wauben-
PIGMENT EPITHELIAL CELL IRON METABOLISM
Penris et al., 1988), and macrophages (Saito et al., 1986; Brock e t al., 1984; Esparza and Brock, 1981) and have, in some cases, been ascribed to a small amount of cell lysis. This is certainly possible since, even after short loading periods, the proportion of total cell 55Fein ferritin i s large. Nevertheless, staining of the cells at the end of the recycling period with trypan blue showed no evidence whatsoever of cell death. This apparent lack of lysis during recycling has also been observed by others who reached the conclusion that lysis could not result in the observed level of secreted iron-binding proteins (Brock et al., 1984). Since some 40% of the “5Fe accumulated by cells in a period of 4 hours is recycled to the environment and about a third of this is the high molecular weight component, this would indicate enough cell death to be readily observed. An alternative explanation is that the high molecular weight material is secreted by the cells without cell lysis as part of the normal iron recycling procedure. It is known that ferritin can be secreted into the serum (Munro and Linder, 1978) and that cardiac cells have ferritin mRNA associated with membrane-associated ribosomes (Campbell et al., 1989). However, serum ferritin is a glycoprotein which binds to concanavalin A (Santambrogio et al., 1987; Cragg e t al., 1981; Wonvood et al., 1979) and we can find no evidence of glycosylation of the high molecular weight ferritin-like material from RPE cells. One piece of evidence supports the contention that this material is not just cytoplasmic ferritin from lyzed cells: It has been found that cytoplasmic ferritin can be immunoprecipitated by a monoclonal and a polyclonal anti-ferritin antibody but the released material cannot. This is not the result of a n inhibitor of antibody binding in the released material since when radiolabeled cytoplasmic ferritin is mixed with recycled material, 55Feis immunoprecipitated in proportion to the cytoplasmic component. It is quite possible that the phenomena described are part of the mechanism of iron transport to the photoreceptor cells since the eye, particularly the aqueous (Tripathi et al., 1990; Yu and Okamura, 1988) and vitreous humor (Hawkins, 1986), contains substantial amounts of apo-Tf. In addition, recent evidence suggests that cells of the neural retina can synthesize W themselves since Tf mRNA has been detected within them (A. Swaroop, personal communication). Current investigations in our laboratory are attempting to determine precisely which cells in the neural retina make apo-Tf. The apo-Tf made within the retina could bind iron released from the apical surfaces of RPE cells and diffuse to the photoreceptor cell inner segments which we have shown to bear TfRs (R.C. Hunt and A.A. Davis, unpublished observations). This mechanism together with the close apposition of photoreceptors to the RPE cells would overcome the problems of diffusion of iron-bearing Tf through the vitreous humor and the layers of the neural retina to the highly metabolic photoreceptor cells. The nature of the high and low molecular weight iron-containing material secreted by human RPE cells remains to be determined; however, our findings of substantial iron release by RPE cells and the binding of this material to apo-Tf suggest a model to explain the transport of iron from the bloodstream across the BRB
109
to iron-requiring cells of the retina. Additional support is lent to this idea by the observations that neural retina cells can make apo-Tf (Hunt and Davis, unpublished observations; Swaroop, personal communication) and that photoreceptors can bind i t (Hunt and Davis, unpublished observations).
ACKNOWLEDGMENTS We thank the South Carolina Lions’ Eye Bank for supplying human eye material for cell culture. This work was supported by grant EY06164 from the National Institutes of Health. LITERATURE CITED Barnes, D., and Sato, G. (1980) Serum-free cell culture: A unifying approach. Cell, 22r649-655. Brock, J.H., Esparza, I., and Logie, A.C. (1984) The nature of iron released by resident or stimulated mouse peritoneal macrophages. Biochim. Biophys. A d a , 797t105-111. Campbell, C.H., Ismail, R., and Linder, M.C. (1989)Ferritin mRNA is found on bound as well as on free polyribosomes in rat heart. Arch. Biochem. Biophys., 273t89-98. Ciechanover, A., Schwartz, A.L., Dautry-Varsat, A., and Lodish, H.F. (1983) Kinetics of internalization and recycling of transferrin and the transferrin receptor in a human heptoma cell line. J. Biol. Chem., 258t968 1-9689. Clark, V.M. (1986) The cell biology of the retinal pigment epithelium. In: The Retina: A Model for Cell Biology Studies. Part 11. R. Adler, and D. Farber, eds. Academic Press, Orlando, Florida, pp. 129-169. Cragg, S.J., Wagstaff, M., and Worwood, M. (1981) Detection of glycosylated subunit in human serum ferritin. Biochem. J., 199t565-571. Cunha-Vaz, J. (1979) The blood-ocular barriers. Surv. Ophthalmol., 23:279-296. Dautry-Varsat, A., Ciechanover, A., and Lodish, H.F. (1983) pH and the recycling of transferrin during receptor-mediated endocytosis. Proc. Natl. Acad. Sci. U.S.A., 80t2258-2262. Esparza, I., and Brock, J.H. (1981) Release of iron by resident and stimulated mouse peritoneal macrophages following ingestion and degradation of transferrin-anti-transferrin immune complexes. Br. J. Haematol., 49t603-614. Fuller, S.D., and Simons, K. (1986) Transferrin receptor polarity and recycling accuracy in “tight” and “leaky” strains of Madin-Darby canine kidney cells. J . Cell Biol., 103t1767-1779. Hanover, J.A., and Dickson, R.B. (1985) Transferrin: Receptor-mediated endocytosis and iron delivery. In: Endocytosis. I. Pastan, and M.C. Willingham, eds. Plenum Press, New York and London, pp. 131-161. Hawkins, K.N. (1986) Contributions of plasma proteins to the vitreous of the rat. Curr. Eye Res., 5:655-663. Hunt, R.C., Ruffin, R., and Yang, Y-S. (1984)Alterations in the transferrin receptor of human erythroleukemic cells after induction of hemoglobin synthesis. J. Biol. Chem.,259t9944-9952. Hunt, R.C., Dewey, A,, and Davis, A.A. (1989)Transferrin receptors of the retinal pigment epithelium are associated with the cytoskeleton. J. Cell Sci., 92:655-666. Jefferies, W.A., Brandon, M.R., Hunt, S.V., Williams, A.F., Gatter, K.C., and Mason, D.Y. (1984) Transferrin receptor on endothelium of brain capillaries. Nature, 312t162-163. Klausner, R.D., Ashwell, G., van Renswoude, J., Harford, J.B., and Bridges, K.R. (1983a) Binding of apotransferrin to K562 cells: Explanation of the transferrin cycle. Proc. Natl. Acad. Sci. U.S.A., 80:2263-2266. Klausner, R.D., van Renswoude,J.,Ashwell, G., Kempf, C., Schechter, A.N., Dean, A., and Bridges, K. (198313)Receptor-mediated endocytosis of transferrin by K562 cells. J. Biol. Chem., 258:47154724. Laemmli, U.K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature, 227t680-685. Lakhanpal, V., Schacket, S., and Rouben, J. (1984) Desferrioxamine (Desferabinduced toxic retinal pigmentary degeneration and presumed optic neuropathy. Ophthalmology, 91t443-451. Munro, H.N., and Linder, M.C. (1978)Ferritin structure, biosynthesis and role in iron metabohsm. Physiol. Rev., 58:317396. Pedersen, 0. (1979)An electron microscopic study of the permeability of intraocular blood vessels using lanthanum as a tracer in vivo. Exp. Eye Res., 21t61-69.
110
HUNT AND DAVIS
Peyman, G.A., and Bok, D. (1972).Peroxidase diffusion in the normal and laser-coagulated primate retina. Invest. Ophthalmol., 21:35-
45. Saito, K., Nishisato, T., Grasso, J.A., and Aisen, P. (1986) Interaction of transferrin with iron-loaded rat peritoneal macrophages. Br. J . Haematol., 62:275-286. Santambrogio, P., Cozzi, A,, Levi, S., and Arioso, P. (1987) Human serum ferritin G-peptide is recognized by anti-L ferritin subunit antibodies and concanavalin A. Br. J. Haematol., 65.235-237. Shichi, H. (1969) Microsomal electron transport system of bovine retinal pigment epithelium. Exp. Eye Res., 8r60-68. Smith, R.S., and Rudt, L.A. (1975) Ocular vascular and epithelial barriers to niicroperoxidase. Invest. Ophthalmol., 14:55&560. Soda, R., and Tavassoli, M. (1984) Transendothelial transport (transcytosis) of iron-transferrin complex in the bone marrow. J . 1Jltrastruct. Res., 88:18-24. Tavassoli, M.: Kishimoto, T., Soda, R., Kataoka, M., and Harjes, K. (1986) Liver endothelium mediates the uptake of iron-transferrin complex by hepatocytes. Exp. Cell Res., 165:369379. Tripathi, R.C., Millard, C.B., Tripathi, B.J., Chailertborisuth, N.S.,
Neeley, K.A., and Ernest, J.T. (1990)Aqueous humor ofcat contains fibroblast growth factor and transferrin similar to those in man. Exp. Eye Res., 50:10%112. Van der Ende, A., Du Maine, A,, Simmons, C.F., Schwartz, A.L., and Strous, G.J. (1987) Iron metabolism in Be Wo chorion carcinoma cells: Transferrin-mediated uptake and release of iron. J. Biol. Chem., 262t8910-8916. Van der Ende, A., Du Maine, A,, Schwartz, A.L., and Strous, G.J. (1989)Effect of ATP depletion and temperature on the transferrinmediated uptake and release of iron by Be Wo choriocarcinomacells. Biochem. J.,259:685-692. Wauben-Penris, P.J.J., Veldscholte, J.,Van der Ende, A., and van der Donk, H.A. (1988)The release of iron by Sertoli cells in culture, Biol. Reprod., 38: 1105-1 113. Worwood, M., Cragg, S.J.,Wagstaff, M., and Jacobs, A . (1979)Binding of serum ferritin to concanavalin A. Clin. Sci., 56.83-87. Yu, T-C., and Okamura, R. (1988)Quantitative study of characteristic aqueous humor transferrin, serum- transferrin and desialized transferrin in aqueous humor. Jpn. J. Ophthalmol., 32:26%274.
