Pathways in the binding and uptake of ferritin by hepatocytes

Biochimica et Biophysica Acta, 1011 (1989) 40-45 Elsevier

40

BBA 12433

Pathways in the binding and uptake of ferritin by hepatocytes K u r t O s t e d o h ~ a n d Philip Aisen L2 I Department of Physiology and Biophysics and : Department of Medicine, Albert Einstein College of Medicine, Bronx, N Y (U.S.A.)

(Received 24 October 1988)

Key words: Ferritin: Ferritin receptor; Iron metabolism; (Rat hepatocyte)

The binding and uptake of rat liver ferritin by primary cultures of rat liver hepatocytes was studied in order to assess the relative importance of saturable, high-affinity pathways and nonspecific processes in the incorporation of the protein by the cells. To minimize artifacts, ferritin not subjected to heat treatment and labeled in vivo with SgFe was used. Binding to cell membranes was estimated from incubations performed at 4 o C. After 2 h, when a steady state in cell-associated ferritin had been achieved, approx. 4 . 1 0 4 binding sites per cell were observed, with an affinity constant for ferritin of 1.109 M - i . At 37°C, the maximal uptake from these sites was 1.3. l0 s ferritin molecules/cell per h. For ferritin molecules bearing an average of 2400 iron atoms, this uptake amounts to 5 . 1 0 6 iron atoms/cell per min. Half-maximal uptake was achieved at a ferritin concentration, or KMI, of 3 . 1 0 - 9 M. Although uptake rates at least a thousand times greater could be achieved by binding to the much larger number of low-affinity sites, the apparent K M, for such 'nonspecific' uptake was 4 • 10-7 M. At ferritin concentrations up to 2 nM, at least 90% of ferritin bound and taken up by hepatocytes involves saturable, high-affinity sites, presumably true ferritin receptors.

Introduction About 30 mg iron is needed for hemoglobin synthesized each day in normal adult man, but only about 1 mg iron per day is provided from nutritional absorption [1]. Thus, nearly all iron needed for erythropoiesis is recycled from senescent and nonviable erythrocytes processed by the reticuloendothelial system [2]. Rat Kupffer cells, the reticuloendothelial macrophages of the liver, phagocytose damaged or immunologicaUy altered erythrocytes [3], and are able to ingest one erythrocyte per cell without obvious damage due to iron overload [4]. The iron derived from catabolized erythrocytes and subsequently released by the phagocytosing cell in vitro can be found in nearly equal amounts either bound to transferrin [5] or incorporated into ferritin [6]. Iron from both proteins is available to the parenchymal cells of the liver (hepatocytes). Although most investigations of transferring uptake into hepatocytes point to the existence of a receptor-mediated mechanism [7-10], other pathways of iron uptake from

Abbreviation: PBS, phosphate-buffered saline. Correspondence: P. Aisen, Department of Physiology and Biophysics, Albert Einstein College of Medicine, 1300 Morris Park Avenue, Bronx, NY 10461, U.S.A.

transferrin have also been incriminated. These include a low-affinity pathway initiated b) ~nonspecific binding of the protein to the hepatocyte membrane (adsorptive endocytosis), fluid phase endocytosis and reductive release of iron at the plasma membrane [9-12]. In contrast, little is known of pathways in the acquisition of iron from ferritin by hepatocytes, although the isolated hepatocyte may obtain iron from ferritin at more than 10-times the rate from transferrin [6,8]. Putative ferritin receptors of the hepatocyte have been described, isolated and partially characterized [13-15], but their role in the uptake of ferritin has not yet been analyzed. In the present study we have sought to evaluate the relative importance of receptor-mediated and nonspecific, nonsaturable pathways in the ferritin-hepatocyte interaction, using rat liver ferritin and primary cultures of rat hepatocytes. Materials and Methods

Preparation of rat liver ferritin A combination of ammonium sulfate precipitation, gel chromatography [16-18] and ultracentrifugation [19] was used to prepare electrophoretically homogeneous ferritin from frozen rat liver (Pel-Freeze Biologicals, Rogers, AR, U.S.A.) or from livers freshly removed from rat carcasses. The commonly used heat-denaturation step was omitted to avoid possible heat-induced changes in the ferritin molecule [20].

01674889/89/$03.50 © 1989 Elsevier Science Publishers B.V. (Biomedical Division)

41 Steps and conditions employed in isolation were: (1) homogenization in the presence of phenylmethylsulfonyl fluoride (50 pM), benzamidine (1 mM), leupeptin (1 pg/ml) and chymostatin (5 #g/ml); (2) lowering pH to 4.8 with acetic acid and subsequent centrifugation for 20 min at 7000 × g; (3) fractional ammonium sulfate precipitation (25%, 45%); (4) ultracentrifugation (2 h at 100000 × g); (5) gel chromatography on Sephacryl S-200 SF (column 1.5 × 60 cm, flow rate 13 ml/h, eluted with phosphate-buffered saline (PBS); (6) concentration by ultracentrifugation for 2 h at 100000 × g. The purity of ferritin was assessed by SDS-polyacrylamide gel electrophoresis using the Laemmli system with a 12.5% separation gel [21]. Ferritin was labeled with 125I to a specific activity of 90000 c p m / p g protein by the Bolton-Hunter method [22]. Ferritin labeled in vivo with 59Fe to a specific activity of 5000-15000 cpm/pg protein was achieved with the procedures of Niitsu et al. [23].

Cultures of primary hepatocytes Hepatocytes were isolated by in situ liver perfusion of male Wistar rats (200-250 g) as described by Berry and Friend [24] and by Seglen [25]. Following isolation, the basis medium for all subsequent steps was RPMI1640 with 2 g / l NaHCO 3 added (GIBCO), containing antibiotics (penicillin 100 U/ml, streptomycin 100 ~g/ml) when used for cell cultivation. The washing media contained 15% newborn calf serum (GIBCO), the culture medium of the first day 15% fetal calf serum (GIBCO) and that of the second day contained a mixture of hormones [26]. Cells were cultured in Linbro multiwell plates. Viability of cells, checked by Trypan blue exclusion after isolation and again on the day of er,periments, exceeded 95%.

Incubation experiments After 2 days of stabilization in culture, the hepatocytes were washed three times with RPMI-1640. Each collagenized tissue culture well [6] was then incubated with 1 ml RPMI-1640 containing 2% bovine serum albumin (Sigma) and labeled ferritin in various concentrations. Binding of ferritin to the collagenized wells has been shown to be negligible [6]. Incubations were stopped by aspirating the media and washing cells four times with PBS, with plates kept cold until scraping. The cells were removed from the wells with a rubber policeman, using 1 ml of 0.1% Triton X-100 in distilled water. The scraped cells and aliquots from the media were counted for radioactivity in a Searle Model 1195 two channel gamma-counter. Protein content of the hepatocyte scrapings was used as a measure of the number of cells present, with 1.28 mg representing 10 6 cells [27]. Cell protein was determined by the Coomassie brilliant stain method [28], with commercial reagent and

standard (Pierce Chemical Co.). This method is not influenced by 0.1% Triton X-100. A control experiment was carried out to determine whether ferritin could release iron to the culture medium during incubation with hepatocytes. Cells were incubated with [59Fe]ferritin at a concentration of 9. 10 -9 M (4/~g/ml), with the incubation medium taken for ultrafiltration at the start of incubation and after 6 hours. At the initial time, 1.0% of the 59Fe was filtrable through an Amicon PM10 membrane filter (molecular weight cutoff, approx. 10000), while at the end of incubation, 1.7% of the labeled iron wzs filtrable. Excluding bovine serum albumin from the culture medium did not affect these results. We conclude, therefore, that release of iron from ferritin to culture medium is negli- C gible.

Analysis of data Binding and uptake parameters were computed by least squares fitting of data using the Simplex algorithm described by Nelder and Mead [29] and a program written in Turbo Pascal. Two classes of binding sites (at 4 ° C) were assumed in the fitting procedures, and best fits obtained with one saturable and the other nonsaturable. In the case of uptake at 37 ° C, best fits were obtained by assuming two saturable processes. Parameters for binding and uptake were obtained from experimental data as follows: At 4°C, a temperature at which internalization of ferritin by hepatocytes is considered negligible, the binding of ferritin to saturable sites on the cell membranes is given by F., =

Bt,,talX k~ X [F] l+k~xlF]

(1)

where F~ represents the number of ferritin molecules bound to saturable sites in each hepatocyte, Bt,,tal specifies total number of saturable binding sites for ferritin per cell, k,, is the association constant for binding of ferritin to strong sites, and IF] is the measured concentration of ferritin free in solution, expressed as concentration in units of molarity. Ferritin bound to nonsaturable membrane sites, F,, is then expressed by Fn=CX[F 1

(2)

with c as the proportionality constant relating nonsaturable binding (molecules per cell) to ferritin concentration. Total bound ferritin, the experimentally measured quantity, is given by Fb,,.nd=

F, + Fn

(3)

Values for BtotaI, k., and c were obtained from the analysis of the experimental binding data of 4 ° C.

42 At 37 ° C, uptake is separated into components representing sites with high and low affinities for ferritin I)hmX IF] /)hi = KMI + [F]

(4)

and ",m X [F] vi° = ~KM2+[F]

(5)

where Vh~and rio are rates of uptake from sites of highand low-affinity for ferritin, respectively; Vhm and /)Ira are corresponding maximal rates of uptake; and KMI and K m are concentrations of ferritin for which halfmaximal rates of ferritin uptake are achieved by sites of high- and low-affinity, respectively (analogous to Michaelis constants). Total uptake is then given by the sum of processes represented in Eqns 4 and 5 t)hm

V'°' =

Vim

K M i + [F~--'--]+ ~ K M 2+ [ E l

)

"[F]

(6)

Using this expression and the experimental values of total uptake at 37 °C as a function of ferritin concentration, corrected for binding at 4 ° C, the parameters Vhm, Vim, KM1 and KM2 are obtained with the simplex curve-fitting algorithm. Results

Binding and uptake of [59Fe]ferritin by hepatocytes Primary hepatocyte cultures were incubated with various concentrations of [SgFe]ferritin for 2 h at 4°C or for 1/2, 2 and 6 h at 37°C (Fig. 1). As indicated, cell-associated ferritin after incubation at 4°C is taken to represent binding to membrane sites [30], while at

37 °C it reflects both binding and uptake or internalization by cells. Since ferritin binding reaches a steady state within 2 h at 4 ° C, data obtained at this time and temperature were assumed to reflect the number of available binding sites at the cell surface. In all experiments, a steep initial rate of increase in cell-associated radioactivity declined, but never reaches zero as the ferritin concentration increased, suggesting easily saturable and much less saturable components in binding and uptake. The linear or nonsaturable phase of ferritin binding continues throughout the concentration range studied, so that this component in binding can be taken as proportional to ferritin concentration. The binding data (2 h, 4°C) are presented as a Scatchard plot in Fig. 2. From this presentation, a class of saturable and a class of nonsaturable membrane binding sites are more readily evident. Each cell displays 41.5.103 saturable binding sites with an association constant of 1.10 9 M - l , and is also capable of binding in a nonsaturable manner 1.74.103 molecules of ferritin for each nM ferritin free in solution. The results of incubation at 37 °C reflect at least two distinct processes: first, binding of the labeled probe to the external surface of the plasma membrane and, second, uptake and internalization by pinocytosis or endocytosis [9]. After 1/2 h at 37°C, the total cell-associated ferritin was barely distinguishable from steady-state binding observed at 4 ° C, suggesting that at least this time is required to saturate membrane binding sites at physiological temperature. The rate of uptake was calculated from the 6 h time point, since uptake has been reported to be linear with time until at least 8 h [6] and the contribution of membrane binding to total cell-associated ferritin is then almost negligible (Fig. 1B). Results are presented in Fig. 3 as an Eadie-Hofstee plot. The curved line can be resolved into two compo-

B

1800 'A 150

¢ 1500 -

~'

12oo -

lOO c

900

"E

0 , 0

~

:

: 5

.. .. .. ..

i .

..

..

,

.

10

Inltiol ferritin concentration

,

.

i

.

. .. . .

15 [ ~j

Z ' . nt ]

, ,

20

.j

600

-

300 ~•

,

.

15

30

45

,, : 60

Initial f e r r i t i n c o n c e n t r a t i o n

:

', ., . . . .! . 75 90

•

,

[ #xj / ml ]

Fig, 1. [SgFelferritin found in hepatocytes after in~'abation with varying concentrations of ferritin at 4 ° C and 37 ° C. The mean of determinations from 1-3 experiments is shown, with each measw:ement performed in triplicate and error bars representing S.D. Where error bars are not shown, the S.D. did not exceed the size of the symbol. 'Cell-associated' refers to the amount of ferritin calculated from radioactivity measured in the washed scraped cells. A, a physiological range of ferritin concentrations: o , 2 h, 4 ° C ; v, 1 / 2 h, 37°C; A, 2 h, 37°C. B, an extended range of ferritin concentrations: o , 2 h, 4 ° C ; i , 6 h, 37°C.

43 80L

=E

c

4o

! E

: f

.~ 40 u

to_ -~x

z° t

i

"i

20.

I o---;-

o.

~

;

o

200 Bound

x

° ....

;

,

10 - 3

400

,

,

o

,

I

0

600

[ m o l e c u l e s p e r cell ]

Fig. 2. Scatchard plot of 59Fe in vivo labeled ferritin bo,und to hepatocytes after incubation at 4 ° C for 2 h. Two classes of binding sites are evident: one saturable (41.5.10 3 sites per cell with an association constant K a of 1.10 9 M - 1), and one non-saturable which binds 1.74 molecules per cell per pM ferritin.

nents: one with a high affinity and a limited maximal uptake and another one with low affinity but an ability to take up large amounts of ferritin. The first components is compatible with receptor-mediated uptake of 7.7.105 ferritin molecules in 6 h, or 1.3.10 3 per h, with half-maximal velocity achieved at 3.15 nM free ferritin concentration. Each binding site in this component could then account for the uptake of three ferritin molecules per h. Assuming the average ferritin molecule bears 2400 iron atoms [6], this corresponds to a maximal rate of 5 . 1 0 6 iron atoms per min into a single cell. The other component observed in ferritin uptake may then be regarded as related to the nonsaturable binding seen during incubation in the cold. This component appears to have some 10-times higher capacity

3.0 -I ~

2.5

i

2.0

~

!.5

t 0

I ' I 10 20 IniUol ferdtln concentration [ / ~ / ml ]

,

I 30

Fig. 4. Binding of ]25I-labeled ferritin and [59Fe]ferritin to hepatocytes after 2 h incubation at 4 ° C. Data points represent the means of determinations in triplicate. B, 125I-ferritin; o, [59Fe]ferritin.

(6.5.10 6 molecules per cell in 6 h, or I • 10 6 per h) than its saturable counterpart, but its apparent Michaelis constant (400 nM) lies beyond the experimentally tested concentration range. Reliable values for maximal uptake by the second component cannot, therefore, be inferred from our experiments. However, such uptake may be irrelevant for conditions in vivo where sufficiently high concentrations of extracellular ferr, fin are not likely to be attained.

Results with/251.1abeled ferritin Results of binding experiments at 4 ° C using 1251labeled ferritin were far less reproducible and not fully in accord with those obtained with [~9Fe]ferritin. The result of one experiment is presented in Fig. 4 together with a plot showing the binding of [59Fe]ferritin. Since some 5-10% of the ]251 label in the ferritin preparations could not be precipitated with 15% trichloracetic acid, despite repeated passage through Sephadex G-25 columns, we believe our studies were confounded by the presence of unbound but unremovable 125I in our preparations. This would account for our finding that some 4 . 1 0 a molecules of ferritin appeared to bind very tightly to each hepatocyte, with a dissociation constant close to zero. We have therefore confined our analyses to data from experiments with ferritin labeled in vivo with 59Fe. Discussion

o

.

0.5 '

c

•ff LL

0

I

o , :1: : 0 I , J ' , I ' ~ : : : : : : : : : " ~ ' r : 0

50

I O0

Uptoke / free x 10

-12

t 50

200

0 : 250

[ molecules per cell / M ]

Fig. 3. Eadie-Hofstee plot of 59Fe in vivo labeled ferritin uptake into hepatocytes after incubation at 37 o for 6 h. There are two pathways of ferritin uptake by the cell: a highly saturable one suggesting a receptor mediated process (Vmax = 7.7" 10 5 molecules per cell in 6 h, or 1.3 molecules/cell per h, KM] = 3-10 -9 M) and a less saturable one suggesting nonspecific uptake (approx. 6.5-10 6 molecules per cell in 6 h, or 1.10 6 molecules/cell per h, apparent K m = 4 " 1 0 -7 M).

Binding of ferritin by hepatocytes In earlier studies by Mack et al. [13], 32.10 ~ ferritin receptors with a binding constant of 1 • 10 8 M-~, were found on each hepatocyte. These studies used a probe of 125I-labeled ferritin at fixed concentration and varying concentrations of unlabeled ferritin for incubations with cells. The number of receptor sites was then calculated from the displacement of labeled ferritin with unlabeled ferritin. Since tb ~. binding data in incubations with hepatocytes did not yield an affinity constant, a

44 putative ferritin receptor was isolated by affinity chromatography and bound to an artificial matrix. The immobilized ferritin receptor was then used for binding studies to obtain an affinity constant. We could confirm that no binding constant can be obtained from experiments with hepatocytes using 125I-labeled ferritin. Moreover, Fig. 4 shows that radioiodinated and in vivo [59Fe]ferritin may give different results. However, we succeeded in obtaining a binding constant with ferritin labeled in vivo with 59Fe and the results are shown in Fig. 1 (direct graph) and in Fig. 2 (Scatchard diagram). The numbers of binding sites per cell estimated with both methods are fairly close (32.103 and 42. 103), but the affinity constant from our experiments, based on direct measurements with hepatocytes, is 10-fold greater than that obtained in the earlier study. This apparent discordance may be due to problems with ]251-labeled ferritin, as discussed above, or to differences in the molecular environment of the receptor when anchored in the original membrane or on an artificial matrix. Both Mack et al. [13] and we believe that the strong binding to a limited number of sites involves a true ferritin receptor. We have also found a second class of more weakly binding, nonsaturable sites which accept 1.7.103 ferritin molecules per cell for each nM ferritin free in solution. In more recent studies, Adams et al. [15] have isolated a fcrritin-binding protein from human liver. This protein, with an apparent subunit molecular weight by SDS-gel electrophoresis of 53000, displays an affinity constant for ferritin binding of 6. l0 s M -1, within a factor of 2 of what we now report. The isolated molecule is presumed to represent the true hepatocyte receptor for ferritin.

Uptake into the interior of the cell Ferritin may be transported into the interior of the hepatocyte by at least three possible paths: (1) receptor-mediated, and therefore saturable, uptake; (2) adsorptive uptake via weak or nonspecific binding sites, and therefore occurring only at high ferritin concentrations; and (3) fluid phase pinocytosis. Since the volume of liquid taken up by the last process is known [30], the amount of ferritin internalized by this pathway can be calculated directly from the free ferritin concentration. This pathway contributes less than 2% to total uptake even at the high end of experimentally tested ferritin concentrations. Therefore, fluid phase pinocytosis can be regarded as insignificant for ferritin uptake, leaving the other two routes to be considered. Two different pathways can be inferred from the uptake data shown in Fig. 3. The first of these, a pathway of limited capacity characterized by a low concentration of ferritin necessary to achieve a halfmaximal rate of uptake (KM]), presumably reflects a specific receptor-mediated process. KM] , the analog of a

A '~

20

s

,= 1 / i

~o~....

:l

.... . .... . .......... o

?

. ....

.....

0

o

x

80

]

eo

.Q

eee Qee

C

~

201

~

o

0

10

20

30

Free ferritin [ nM ] Fig. 5. (A) Expected ferritin binding at 4 ° C by hepatocytes, calculated from parameters shown in the Scatchard plot of Fig. 2 for binding of [59Fe]ferritin. (B) Binding over a wider range of ferritin concentration to demonstrate receptor saturation at concentrations exceeding the physiological range ( > 2 nM). - - - - - , binding to readily saturable site; . . . . . . , binding to 'nonspecific' site; ~ , total binding.

Michaelis constant, does not exactly match the binding constant of the saturable class of binding sites. However, a binding constant is determined at a stationary equilibrium, while the Michaelis constant is derived from measurements of the rate of uptake, a process that may not reflect true thermodynamic equilibrium in binding of ferritin to its receptors in the metabolizing cell. The other pathway, of much greater capacity than the first, requires a concentration of ferritin to attain half-maximal velocity that is not reached in our experimental conditions, and is well beyond concentration ranges observed in the circulation [31]. This pathway may be regarded as entailing weak binding to a large number of nonspecific adsorption sites. The experimen tally determined parameters for the binding and ,- 5 of ferritin by hepatocytes were used to estimate tat relative contributions of the two classes of sites in binding and uptake. First, ferritin bound to each class of binding sites, saturable (strong) and nonsaturable (weak), was calculated for the range of ferritin concentrations encountered in the circulation in normal and pathologic states (Fig. 5A and B). Then, uptake from each class was calculated (Fig. 6A and B). In both instances, weak binding sites contributed little to total binding or total uptake within the range of ferritin

45 ,-..,

80

,,

This work was supported in part by Grant DK3792"r from the National Institutes of Health, U.S. Public Health Service.

60 o.

40

o

20

Acknowledgements

References

0 I o

J 0 .... B

X

c

"E I

,t.

o~

"G ~. =

1

2

3

"

4 '

o 5

. . . .

200.

"'

150. e

100.

,, ,--

....

5o. 0 ~-~-" 0

"---~

: 20 Free ferritin

! 40

,

; 60

•

[ nM ]

Fig. 6. (A) Expected ferritin uptake at 37 o C by hepatocytes, calculated from parameters shown in the Eadie-Hofstee plot of Fig. 3 fnr uptake of [SgFe]ferritin. (B) Uptake over a wider range of ferrit .a concentration to demonstrate involvement of the low-affinity 'nonspecific' pathway. - - - - - , uptake via saturable pathway; . . . . . . , uptake via low-affinity pathway; ~ , total uptake.

concentrations encountered in the circulation. Only when the concentration of ferritin accessible to the hepatocyte exceeds 2 nM (900 /tg/l), or when more than 2/3 of the saturable sites are occupied, does uptake make appreciable use of the pathway of low-affinity, nonsaturable binding. All of our studies have been carried out with nonglycosylated tissue ferritin. Whether they can be extrapolated to the interaction of hepatocytes with circulating ferritin, which is largely glycosylated [32,33], is not now clear. Further, nothing is known of the concentration of nonglycosylated ferritin in the extracellular fluid bathing hepatocytes in vivo, so we can make no inferences about the physiological role of ferritin as an iron donor. Our results on the interaction of ferritin with hepatocytes may be contrasted with those reported earlier for transferrin and hepatocytes [11]. As with ferritin, transferrin is bound to strong, saturable sites as well as to more plentiful but more weakly binding sites. At less than physiological concentrations, binding of the protein to the cells, and uptake of its iron, involves predominantly the strong sites, thought to be specific transferrin receptors [8,9]. However, at physiological concentrations of transferrin, the weaker binding sites may contribute more to overall binding and uptake, presumably by a process of adsorptive endocytosis, and fluid-phase endocytosis may also become important [9]. Comparable nonspecific pathways do not seem to contribute substantially to the interaction of ferritin with hepatocytes.

1 Forth, W. and Rummel, W, (1973) Physiol. Rev. 53, 724-792. 2 Noyes, W.D,, Bothwell, T.H. and Finch, C.A. (1960) Br. J. Haematol. 6, 43-55. 3 Munthe-Kaas, A.C. (1976) Exp. Cell Res. 99, 319-327. 4 Kondo, G., Saito, K., Grasso, J.A. and Aisen, P. (1988) Hepatoiogy 8, 32-38. 5 Saito, K.. Nishisato, T., Grasso, J.A, and Aisen, P. (1986) Br. J. Haematol. 62, 275-286. 6 Sibille, J.-C., Kondo, G. and Aisen, P. (1988) Hepatology 8, 296-301. 7 Young, S.P. and Aisen, P. (1980) Biochim. Biophys. Acta 633, 145-153. 8 Young, S.P. and Aisen, P. (1981) Hepatoiogy 1, 114-119. 9 Trinder, D., Morgan, E. and Baker, E. (1986) Hepatology 6. 852-858. 10 Thorstensen, K. and Romslo, !. (1988) J. Biol. Chem. 263, 8844-8850. l l Page, M.E., Baker, E. and Morgan, E.H, (1984) Am. J. Physiol. 246, G26-G33. 12 Thorstensen, K. and Romslo, !. (1984) Biochim. Biophys. Acta 804, 393-397. 13 Mack, U., Powell, L.W. and Halliday, J.W. (1982) J. Biol. Chem. 258, 4672-4675. 14 Mack, U., Storey, E.L., Powell, L.W. and Halliday, J.W. (1985) Biochim. Biophys. Acta 843, 164-170. 15 Adams, P.C,, Powell, L.W. and Halliday, J.W. (1988) Hepatology 4, 719-721. 16 Munro, H.N. and Linder, M.C. (1978) Physiol. Rev. 58, 317-396. 17 Linder, M.C. and Munro, H.N. (1972) Analyt. Biochem, 48. 266-278. 18 Harrison, P.M., Clegg, G.A, and May, K. (1980) in Iron in Biochemistry and Medicine, !!. (Jacobs, A. and Worwood, M., eds.), pp. 131-171, Academic Press, London. 19 Penders, T.J., DeRooij-Dijk, H.H. and Leijnse, B. (1968) Biochim. Biophys. Acta 168, 588-590. 20 Frenkel, E.J., Van den Beld, B., Van Oost, B.A. and Marx, J.J.M. (1983) Biochim. Biophys. Acta 745, 202-208. 21 Laemmli, U.K. (1970) Nature 227, 680-685. 22 Bolton, A.E and Hunter, W.M (1973) Biochem. J. 133, 529-539. 23 Niitsu, Y., Adashi. C., Takahashi, F., Goto, Y., Kohgo, Y., Urushizaki, I. and Listowsky, I. (1985) Am. J. Haematol. 18, 363-371. 24 Berry, M.N. and Friend, D.S. (1969) J. Cell Biol. 43, 506-520. 25 Seglen, P.O. (1976) in Methods in Cell Biology. Voi. Xlll (Prescott, D.M., ed.), pp. 29-83, Academic Press, New York. 26 Reid, L. (1988) in Methods in Molecular Biology: Vol 5: Tissue Culture (Pollard, J.w. and Walker, J.M., eds.), in press. 27 Sibille, J.-C., Octave, J.-N., Schneider, Y.-J., Trouet, A, and Crichton, R. (1986) Eur. J. Biochem. 155, 47-55. 28 Bradford, M.M. (1976) Analyt. Biochem. 72, 248-254. 29 Neider, J.A. and Mead, R. (1965) Comput. J. 7, 308-313. 30 Ose, L., Ose, T., Reinertsen, R. and Berg, T. (1980) Exp. Cell Res. 126, 109-119. 31 Drews, J., Diillmann, J., Hausmann, K. and Kuse. R. (1980) in Serumferritin: Methodische und Klinische Aspekte (Kaltwasser. J.P. and Werner, E., eds.), pp. 194-205, Springer-Verlag, Berlin. 32 Halliday, J.W., Mack, U. and Powell, L.W. (1979) Br. J. Haematol. 42, 535-546. 33 Cragg, S.J., Wagstaff, M. and Worwood, M. (1980) Clin. Sci. 58, 259-262.