Britirhjournal oftiaematology, 1980,45,607420.

The Regulation of Iron Release from the Perfused Rat Liver E. BAKER, A. G. MORTON* AND A. S . TAV1LL.t

Division of Clinical Investigation, Clinical Research Centre, Harrow, Middlesex, and Department ofExperimenta1 Haematology, University College Hospital Medical School, London

(Received 4 October 1979; accepted for publication 1 February 1980)

SUMMARY. Factors affecting iron efflux from the isolated perfused rat liver were studied following the intravenous administration of tran~ferrin-~~Fe or transfer- r i r~ -~~Fc administered to the rat from 1.5 h to 3-5 d before perfusion of the liver.

The liver was perfused with rat red cells suspended either in rat plasma or Eagle’s Basal Medium (EBM). The mean rate ofefflux into a plasma pool containing normal iron and transferrin concentrations was 0.9% of the initial hepatic radioactive iron pool per hour. In EBM the average rate of efflux was O-l%/h and this could be increased to the rate observed with plasma by the addition of apotransferrin. The rate of iron release from the liver in the presence of apotransferrin or other chclators was inversely proportional to the time of prelabelling. Maximal release rates were observed in livers perfused within 5 h of administering tran~ferrin-~~Fe to the rat. The effect of apotransferrin on efflux into EBM was concentration dependent. However, the maximum release of liver iron by apotransferrin occurred at physiolo- gical apotransferrin concentrations and addition of apotransferrin to plasma pro- duced no increase in the rate of iron ef€lux. The stimulation of iron release in EBM caused by apotransferrin could be reversed by reducing the unsaturated iron binding capacity of the perfusate, either by addition of iron or removal of apotransferrin. However, increasing the iron concentration in the pcrfusate by the addition of iron-saturated transferrin without any reduction in the unsaturated iron binding capacity additionally increased iron release into plasma and EBM. This prcsumably reflects the exchange of plasma transferrin-”Fe for liver 59Fe. Hence iron release measured in these studies reprcscnts the sum of two processes-net release of 59Fe induced by apotransferrin and iron exchange between plasma and liver iron pools.

Apotransferrin and desferrioxamine were equally effective, per unit iron binding capacity, in mobilizing liver iron, and may compete for the same parenchymal iron pool. This suggests that mobilization of iron by apotransferrin may depend solely on

Present address: Protein Fractionation Centre, Scottish Blood Transfusion Service, Ellen’s Glen Road, Edin-

t Present address: Department of Medicine, Case Western Reserve University at Cleveland Metropolitan

Correspondence: Dr E. Baker, Department of Physiology, University of Western Australia, Nedlands, Western

burgh EH17 7QT.

General Hospital, 3395 Scranton Road, Cleveland, Ohio 44109, U.S.A.

Australia 6009.

0 0 0 7 - 1 0 4 8 / 8 0 / ~ 7 $ 0 2 . 0 0 0 1980 Blackwell Scientific Publications

607

608 E. Baker, A . G. Morton and A. S . Tavill its ability to chelate ferric iron and not on a more specific ferroxidase activity or interaction with membrane receptors.

The mechanisms involved in the transport of iron across the liver cell membrane and the regulatory factors controlling this exchange are not known. Liver iron uptake may involve the reversible binding of transferrin to specific receptor sites (Gardiner & Morgan, 1974) in a process similar to that in erythropoietic cells (Jandl & Katz, 1963) and placenta (Laurel1 & Morgan, 1964). Net transfer to the hepatocyte is related to the iron stores in the liver and may be regulated in part by the available storage capacity of apoferritin (Morton & Tavill, 1978). The regulation of liver iron mobilization is poorly understood. Caeruloplasmin, xanthine oxidase, ascorbic acid, the plasma iron and transferrin levels and the naturally occurring iron chelates citrate and fructose may be involved in iron rqlease from the liver. The relative importance of these compounds is not certain, and is complicated both by species variations in their activity and by differential effects on parenchymal and reticuloendothelial cells.

The apparent differences in iron release from reticuloendothelial cells and hepatocytes probably reflect the variable forms and distribution of iron in these cells. In the rat almost all liver ferritin iron is localized in the hepatocyte (Van Wyk et al, 1971) while iron in the reticuloendothelial cells is largely in a non-haem, non-ferritin form, and may be partly in the ferrous state (Lipschitz et al, 1971). Studies using a quantitative autoradiographic technique (Hershko et al, 1973) have shown that the reticuloendothelial cells and hepatocytes can be selectively labelled with different forms of radioactive iron. The use of these markers in the isolated perfused liver makes it possible to identify reticuloendothelial and parenchymal iron exchange in isolation from other body iron pools. In the present study iron release from the perfused rat liver has been measured following labelling of hepatocyte iron with the parenchy- mal marker transferrin-bound radioactive iron. Parts of this study have been presented a t the Second International Symposium on the Proteins of Iron Storage and Transport (Baker et aJ, 1975).

METHODS

Human transferrin was obtained from Behringwerke, A.G. (Marburg-Lahn, Germany). Rat transferrin was purified using the method of Gordon & Louis (1963). All chemicals used were analytical grade quality. Bovine serum albumin (Fraction V, 9699% albumin) was obtained from Sigma Chemical Co. (St Louis, Missouri, U.S.A.), Desferrioxamine (DFO) was obtained from CIBA (Desferal, Basel, Switzerland), and Amicon filters from Amicon (Lexington, Massachusetts, U.S.A.) . Iron-59 (FeCls, 5-15 pCi/,ug) and iron-55 (FeCl3, 200-400 pCi/pg) were obtained from The Radiochemical Centre (Amersham, Bucks, England).

Labelling of the Liver in Viuo Rat plasma, purified transferrin and purified rat human transferrin were labelled with 59Fe or

55Fe using a 20 molar excess of citrate (Bates et al, 1967) 1 d before administration. The label was injected intravenously into %month-old, male Sprague-Dawley rats fed a standard pelleted diet ad libitum. The amount of iron taken up by the liver (1-1.5 pg) represents trace labelling with radioactive iron, since the total hepatic iron content was over 1 mg.

Iron Releasefrom Perjiised Rat Liver 609

Liver Perjiision Liver perfusions were performed as described previously (Morton & Tavill, 1977, 1978).

The liver was washed free of labelled blood by flushing with 50 ml cold perfusate immediately before the perfusion. The perfusate consisted of washed rat red cells suspended (PCV 30%) in either heparinized rat plasma or Eagle’s Basal Medium (EBM; Eagle, 1955) containing bovine serum albumin (28 g/l) and buffered to pH 7.4 with sodium bicarbonate (0.01 M) and Hepes (0.02 M). The liver was perfused in an oxygenated, recycling system with a known volume of perfusate and “Fe release was measured in perfusate samples withdrawn at 15-30 min intervals. Apotransferrin, transferrin and iron concentrations in the perfusate were adjusted by adding apotransferrin, iron-saturated transferrin, or ferrous ammonium sulphate dissolved in 9 g/1 NaCl (pH 5) immediately before use.

At the end of the perfusion, the liver was removed, perfused with cold isotonic saline, and homogenized in an equal volume ofice-cold saline. Duplicate 1 ml samples of the homogenate, perfusate sample supernatants and bile were counted in an automatic gamma spectrometer with standards of the injection mixture. In experiments in which the liver had been doubly labelled with 55Fe and 59Fe, ”Fe was estimated by liquid scintillation counting after decay of the 59Fe, or simultaneously using the procedure of Harris & Aisen (1975), except for liver homogenates which were treated with increased concentrations of Clorox and ascorbic acid and counted at 5°C after 48 h. The net accumulation of radioactive iron in the perfusate was calculated after appropriate corrections for change in volume and for radioactivity removed by sampling. The initial hepatic iron radioactivity was calculated from the sum of radioactive iron in the liver homogenate, the perfusate samples and the bile.

Analytical Methods Plasma iron and total iron binding capacity (TIBC) were measured using an adaptation

(Morton & Tavill, 1978) of the method of Young & Hicks (1965). Total liver iron was measured using a modification (Morton & Tavill, 1978) of the method of Tsao & Belliles (1 973). Ferritin iron concentration and ferritin radioactivity were measured after isolation by an adaptation (Linder & Munro, 1972) of the method of Drysdale & Munro (1965). Hepatic non-ferritin iron and radioactivity were calculated by difference.

Statistics The rate of efflux of iron was measured by least mean squarc linear regression of the total

perfusate radioactivity values. The significance of the differences in data was tested using the Student’s t test.

RESULTS

Iron release from the rat liver was measured after prelabelling iti vivo with radioactive iron 1.5 h to 3.5 d before the perfusion. Radioactive iron was released from the perfused prelabelled liver a t linear rates when perfused initially with plasma (0-3 h) and then with EBM (3-6 h) (Fig 1). A similar decrease in rate of iron release was observed in two other experiments when the perfusate was changed from plasma to EBM. Liver cell death, reflected in a sharp increase in release ofS9Fe and lactic dehydrogenase was usually observed a t 6-8 h. The marked, irreversible

610 E. Baker, A. G. Morton and A. S . Tavill

't

l-

r i d FIG 1 . The release of 59Fe from a rat liver perfused initially with normal plasma followed by Eagle's Basal Medium (EBM). Release of 59Fe is expressed as % of the initial hepatic radioactivity.

increase in iron and enzyme release due to cell death was easily distinguishable from the smaller reversible changes in iron efflux due to experimental manoeuvres.

Efects of Time of Relabelling Both the amount of radioactivity taken up by the liver and the rate of efflux were affected by

the time of prelabelling. As shown in Fig 2A the proportion of intravenous transferrin 59Fe or 55Fe taken up by the liver increased rapidly in the first few hours after injection, reached a maximum of about 9% of the dose within 1 d and then showed a tendency to decline between 1.0 and 2.5 d. The rate of release of liver radioactivity into a plasma perfusate was greater in the first few hours after prelabelling, but decreased to a steady lower value for prelabelling times between 0.5 and 3.0 d (Fig 2B). The release of iron into EBM containing desferrioxamine was affected in a similar manner by the time of prelabelling (Fig 2B).

EJect of T y p e ofPe$usion Medium The rate of 59Fe efflux was markedly affected by the perfusate medium. Table I shows the

iron content and kinetics of iron release in livers perfused with either plasma or EBM after prelabelling with tran~ferrin-~~Fe between 0.5 and 3.5 d before the experiment. The mean rate of iron efflux into a plasma perfusate was 0-92%/h of the initial hepatic radioactive iron pool. This was significantly higher than the rate of efflux into EBM (0*13%/h). The rate of 59Fe release into bile was 0.17&0.06%/h (mean+SEM, n = l l ) in livers perfused with plasma and 0.14f0.05%/h (n=4) in livers perfused with EBM. The difference between these rates was not significant.

Iron Release-from Perjiused Rat Liver A

61 1

FIG. 2. Effect ofprelabelling time on net liver uptake ofs9Fe or 55Fe-transferrin (A) and on the net rate of release of liver radioactivity (B). Efflux was measured in normal plasma (Unsaturated Iron Binding Capacity, UIBC=55 pt; o), or in EBM (D), or in EBM containing desferrioxamine (UIBC=55 p ~ ; x), and expressed as % of liver radioactivity/h.

TARLE I. Iron stores and kinetics in rat livers perfiused with plasma or EBM after prelabelling in vim with tran~ferrin-~~Fe*

Perfirsed with Perjiisrd with plmrna EBM (n=12) (n=8) P

Liver uptake ofj9Fe (% dose) 6.57k0.72-t 8.26k 1.38 NS Liver weight (g) 15.0k0.7 14.5k0.7 NS Total hepatic iron (pmol/lO g liver) 11.91+0.55 13.24k1.84 NS Ferritin hepatic iron (pmol/lO g liver) 3.44f 0.40 4.37k0.73 NS Rate of s9Fe release (% hepatic 59Fe/h) 0.92k0.18 0.13k0.04 <0.01 Absolute release rate (pmol Fe/lO g liver/h) calculatcd from: (i) Total hapetic iron 0.105+.0.020 0~016+0~OM <0.01 (ii) Ferritin hepatic iron 0.026 0.005 0 . 0 6 0.003 < 0.01 (iii) Total hepatic iron specific activity 0.168+0.052 0-031 kO.011 =0.05 (iv) Ferritin hepatic iron specific activity 0.105k0.023 0.033 k0.012 ~ 0 . 0 5

* Experiments with prelabelling times less than 0.5 d were not includcd. t Mean k SEM.

Four estimates of the absolute rate of hepatic iron release into the perfusate were calculated from the fractional rate of efflux and: (i) the total liver iron pool; (ii) the ferritin iron pool; (iii) the total liver iron specific activity; (iv) the ferritin iron specific activity. The estimatcd absolute rate of iron release varied with the iron pool selected for the calculation. Hershko et al (1973) found that 97-100% of the iron taken up by the liver from transferrin was localized in

612 E. Baker, A . G. Morton and A, S . Tavi l l

Time of Rsrfusion (hours)

FIG 3

ApoTf added

FIG 3. The effect on the hepatic release of 59Fe into EBM of adding apotransferrin (1.1 mg/ml of perfusate) and sufficient iron (0.03 p / m l of perfusate) to saturate the apotransferrin.

FIG 4. The effect of apotransferrin concentration on 59Fe release from rat livers perfused with EBM after prelabelling 0-15 d (0) or 0.77 d (0) previously. Apotransferrin aliquots of 1.5 mg/ml were added to give successive total perfusate concentrations of 1.5,3.0 and 4-5 mg/ml.

the hepatocyte. If this formed a homogenous pool in ferritin, which is almost entirely localized in the hepatocyte in the rat (Van Wyk et al, 1971), the best estimate of the absolute rate ofiron release would be obtained using the measured ferritin iron pool. The actual rate of iron release into plasma calculated in this manner was 0.03 pmol/lO g liver/h (Table I) using livers prelabelled more than 0.5 d prior to the perfusion. This rate is similar to published values for the rate ofiron release from liver ferritin in v i vo (Unger & Hershko, 1974). The absolute rate of iron efflux into EBM calculated using the ferritin iron pool was 0-006 pmol/lO g liver/h (Table I), significantly lower than that observed using a plasma perfusate. Higher estimates of the absolute rates of iron efflux into plasma and EBM were obtained when iron release was related to total liver iron or to total iron or ferritin specific activities.

E#ect of Addition of Aptransferrin The rate of radioactive iron release into EBM was increased to a level near that in plasma by

the addition of apotransferrin (Fig 3) . In 10 experiments the mean rate of iron release increased from 0.12+0.03%/h (mean &SEM) to 0.80f0.15%/h of the hepatic radioactive iron pool when apotransferrin (1.1-1.5 mg/ml) was added to the perfusate. This increase was highly significant (P<O-oOl). The effect of apotransferrin on hepatic iron release into a plasma

Iron Release from Peerftrsed Rat Liver 613

perfusate was studied in a similar manner. In five experiments the mean rate of iron release changed from 0.78+0.17%/h to 0.64 & 0.20%/h when apotransferrin (1.1-1.5 mg/ml) was added. This difference was not significant.

The effect of apotransferrin on liver iron mobilization in EBM was concentration dependent (Fig 4). The rate of iron release increased with increase in apotransferrin concentration up to a maximum rate similar to that achieved with plasma. A5 shown in Fig 4, efflux in EBM containing apotransferrin was affected by prelabelling time in the same manner as efflux into plasma (Fig 2B). The rate of efflux in EBM containing 3 mg/ml apotransferrin was much greater after a short prdabelling time (3.9%/h a t 0.15 d), than after a long prelabelling time (0-6%/h a t 0.77 d). These rates are similar to those seen in plasma at the same prelabelling times (Fig 2B).

To determine whether the inorganic iron and protein impurities detected in the commercial preparation of bovine serum albumin affected the release of iron from'the liver, a comparison was made of release rates measured in the presence of the commercial albumin and purified albumin. The albumin was purified by dialysis against EDTA (0.001 M), chromatography on G150, dialysis against sodium perchlorate (0.01 M), repeated dialysis against phosphate-buf- fered saline and finally against EBM. There was no apparent difference in iron release from livers perfused with EBM containing the commercial or purified albumin (Table 11).

To remove any chelate contaminants introduced during transferrin purification, human or rat apotransferrin (Table 11) was dialysed against perchlorate, saline and repeated changes of deionized water (Price & Gibson, 1972). No change was observed in its effect on liver iron releasc. Human and rat apotransferrin were equally effective in stimulating liver iron release. In three experiments in which rat apotransferrin was added, the rates ofiron release were similar to those observed in seven experiments using human apotransferrin at a similar concentration (Table 11).

Efect of Removing Apotransferrin As shown in Fig 3, the increase in liver iron release produced by apotransferrin could be

reduced by saturation of the circulating transferrin. In each of six experiments where the

TABLE 11. The effect on the rate of efflux of 59Fe from the perfused rat liver ofhuman and rat apotransferrin, dialysed apotransferrin and commercial or purified albumin

Perfusate composition 5yFe release (% hepatic ''Felh) P

Apotransferrin Albumin Basal rate Added apotransferrin (1.1-1.5 mglml)

Human (n=S) Commercial 0.09+ 0.04 0.70 f 0.24 0.05 Human (n = 3) Purified 0.12 (0.18, 0.04, 0.14) 0.88 (1-18, 0.45, 1.02) -

Dialysed rat (3) Commercial 0.21 (0.26,0.15,0.22) 1.02 (0-96.0.66, 1.45) ~-

Dialyscd human (n=2) Commercial 0.11 (0.00, 0.22) 1.08 (0.71, 1.45) -

Results are expressed as mean? SEM with number of experiments in parenthesis, except for groups of two or three, where individual data are given.

614 E. Baker, A. G. Morton and A. S . Tavill

unsaturated iron binding capacity of the perfusate was reduced either by adding iron (as ferrous ammonium sulphate freshly dissolved in isotonic saline at pH 5) or by replacing the perfusate with a transferrin free medium, there was a decrease in the rate of iron release from the perfused rat liver. The mean rate ofhepatic iron efflux decreased significantly (P<0.05) to 29+ 13% of the initial value. However, efflux was not eliminated indicating that a component of iron release was due to a factor other than net chelation by apotransferrin. Two factors considered were release of iron as hepatic ferritin-59Fe and turnover of perfusate transferrin-%Fe with tran~ferrin-~~Fe in a hepatic pool.

Evidence was obtained by chromatography on Sephadex G200 that this background efflux of radioactivity was in the form of a high molecular weight component, possibly ferritin. Since for each mole of iron released by the liver the transferrirxferritin molar ratio is at least 100 (based upon an average content of 225 atoms of iron per molecule of serum ferritin (Zuyder- houdt et al, 1978), this signifies relatively small quantities of hepatic ferritin efflux. However, this may represent a component of the physiological mechanism for hepatic iron release (Siimes & Dallman, 1974; Worwood, 1979) or may be in part due to cell damage.

Effect $Change in Perfusate Iron Content While apotransferrin addition had no significant effect on the rate of 59Fe release from livers

pcrfused with plasma, addition of iron-saturated transferrin increased the rate of 59Fe efflux (Fig 5A). In each of five experiments the rate of iron efflux increased when the plasma iron content of a plasma perfusate was increased by the addition of iron-saturated transferrin or by partial replacement of the perfusate with a plasma aliquot with a higher iron content (Fig 5B). This enhancement of efflux occurred even when there was no change in the unsaturated iron

FIG 5. (A) The effect on the release ofS9Fe into plasma ofadding apotransferrin (apo Tf, 1.5 mglmlof perfusate) and iron-saturated transferrin (Fe Tf, 1.5 mg/ml of perfusate. (B) The rate of iron efflux into plasma before (0) and after (A) increasing the perfusate plasma iron concentration.

Iron Release j o i n P e y h e d Rat Liver 615 Fe to Fe to

~~~ ~

D 1 2 3 4 5

Time of perfusion (hours)

FIG 6. The effect on 5'Fe release into a plasma perfusate of successive additions of: (i) sufficient iron to saturate the plasma UIBC; (ii) wfficient apotransferrin to provide a normal UIBC; (iii) iron to saturate the added apotransferrin.

binding capacity of the plasma. An increase in the rate of "Fe efflux from 0.71 to 1.07%lh was also observed in one perfusion using EBM when iron-saturated transferrin (1.5 mglml) was added to a perfusate containing apotransferrin (1.5 mgiml). This suggests that while a component of liver 59Fe efflux represents a net mobilization of livcr iron, part of the 59Fe release may reflect turnover between circulating transferrin-Fe and the intrahcpatic iron pool.

Efflux of "Fe could be blocked, a t least temporarily, by the addition of sufficient ferrous iron to reduce the unsaturated iron binding capacity to zero by saturation of all transferrin iron-binding sites (Figs 3 and 6). In some experiments (e.g. Fig 6), efflux began to increase again after blocking the binding sites on transferrin and before addition of apotransferrin. Presum- ably the liver removes some of the iron presented to it as transferrin-Fe, and the resultant free apotransferrin iron-binding sites facilitate iron mobilization.

Possibly both iron mobilization and iron exchange require the presence of some free transferrin iron binding sites. While an increase in the apparent rate ofiyFe efflux was observed in individual experiments after increasing the concentration of iron-saturated transferrin in plasma (Figs 5A and 5B), when all the experiments performed with a plasma perfusatc were considered there was no significant correlation between the initial rate of 5'Fe efflux and the starting plasma iron concentration, iron binding capacity or iron saturation. This may be due to the variation between experiments in prelabelling time, dose of iron and liver iron content.

616 E . Baker, A . C . Morton and A . S . Tavi l l

/ .i .--.-'

0 1"'" OwlTf

*/- *-a b-* 1

O O d 5 1 2 3 a 0 ' 0 1 2 3 a 5

Timed perfusion (hours) Timed perfusion (hwrs)

FIG 7. The effect of thc iron chelators apotransferrin (apo Tf) and desferrioxamine (DFO) on 59Fe release from the perfused liver. DFO was added to give a final perfusate UIBC of 54 PM. The prelabelling times for each experiments were 0.13 d (A) and 0.79 d (B).

Efect of Desfrrioxamine To determine the specificity of the apotransferrin induced iron release it was compared with

iron release induced by the ferric chelator desferrioxamine. Apotransferrin and desferrioxa- mine appeared equally effective, per unit iron-binding capacity, in mobilizing liver iron (Fig 7). As shown in Fig 7A, desferrioxamine added to an EBM perfusate 2 h after the addition of apotransferrin at a physiological iron-binding capacity produced no further increase in the rate of iron efflux. Similarly, apotransferrin added to an EBM perfusate in the presence of desferrioxamine had no effect on the rate of iron efflux (Fig 7B). This suggests that apotransfer- rin and desferrioxamine compete for the same hepatic iron pool and that mobilization of iron by apotransferrin may depend solely on its ability to chelate ferric iron rather than a specific interaction with hepatocyte membrane receptors. However, desferrioxamine must operate in part by a different mechanism from apotransferrin. In one experiment in which release of hepatic "Fe into perfusate and bile were measured separately, the addition of desferrioxamine to a plasma perfusate with a normal unsaturated iron binding capacity caused a marked increase in biliary 59Fe excretion.

DISCUSSION

In this study we have used tran~ferrin-~~Fe to label the liver in uivo before liver perfusion. Hershko et a1 (1973) have shown that 97-100% of the tran~ferrin-~~Fe label is localized in the hepatocyte after intravenous administration, and accordingly 59Fe efflux measured in the present experiments was presumed to originate in the hepatocytes. There is no evidence for reticuloendothelial uptake of transferrin-iron in uivo, although iron distribution in the liver may be affected by the iron saturation of the transferrin and the total quantity of iron administered.

In the present study, 7-10%0 of the dose was present in the liver within 1-2 d of injection.

Iron Releasefrom Perjiised R u t Liver 617

This level of incorporation is similar to values published earlier (Gardiner & Morgan, 1974; Milsom & Batey, 1978). The rate of j4Fe release from the perfused liver was much greatcr in the first few hours after labelling than a t longer time intcrvals. This time dependence was observed whether iron release was measured in plasma (Fig 2B), or in EBM containing apotransferrin (Fig 4), or desferrioxamine (Fig 7), but not in the absence of transferrin (Fig 2B). While this data is in agreement with the ‘last in, first out’ hypothesis ofiron release from ferritin (Hoy e t a / , 1974), it is also compatible with the concept ofa chelatable hepatic iron pool (Jacobs, 1977), which represents an intcrmediate iron form between plasma iron and liver ferritin iron. The stabilization of radioactive irqn rclease aftcr about 5 h probably reflects incorporation of most of the liver radioactivity into ferritin. This is in keeping with the time of maximum induction of apoferritin synthesis and incorporation of iron into ferritin seen after administ- ration of iron to both normal and iron deficient rats (Drysdale & Munro, 1966; Morton & Tavill, 1978). After-about 5 h of prelabelling the rate of iron release bccomes constant and presumably rcpresents the basal level of net iron efflux from the liver stores to the plasma. The rates of inorganic iron release calculated using this rate and the liver ferritin contcnt (Table I) are in agreement with published values (Cook et al, 1973; Unger & Herschko, 1974). Howcvcr, in terms of total body fluxes, the rate of iron efflux from the liver parenchyma is low, representing less than 5% of the daily plasma iron turnover in the rat (Cook et al , 1973).

Iron efflux from the perfused liver into a medium free of transferrin was very low or zero, and was markedly increased by apotransferrin to a level near that in plasma, suggesting that apotransferrin is normally required for iron release from the hepatocyte. In most experiments iron continued to be released a t a low rate in the absence of added transferrin. This may have been due to the presence of low molecular weight iron chelates such as amino acids in the EBM or to the release of f~rritin-’~Fe from damaged cells or as part of a normal physiological process (Siimes & Dallman, 1974).

The relationship between apotransferrin concentration and increase in rate of iron releasc in EBM (Fig 4) suggests that iron efflux from the pcrfused liver is maximally stimulated by a physiological level of apotransferrin. The observation that increasing the unsaturated iron- binding capacity of a plasma perfusate by addition of apotransferrin did not further increase iron efflux (Fig 5A) suggests that there is a maximal transport rate which cannot be exceeded by adding more transferrin. This contrasts with the in vivo studies of Hallberg t(i Solve11 (1960) and Lipschitz et al(1971) who observed an increase in plasma iron concentration on administration of apotransfcrrin. This increase was not derived from reticuloendothelial cells. Transferrin may have a regulatory role in iron mobilization, although other rate-limiting proce.sses must be involved which prevent abnormally high rates of iron mobilization in thc steady state, but which facilitate the increased iron release observcd under conditions of increased iron need (Cook et al, 1973).

The increase in iron release from the livcr produced by transferrin was directly related to its iron-binding capacity. The stimulation of iron efflux by apotransferrin was reversed when the iron saturation of the transferrin was increased, or the perfusate replaced with a medium with no unsaturated iron-binding capacity. Transferrin may function simply by providing a receptor for iron in the plasma, thus maintaining a concentration gradient for iron across the liver cell membrane, or it may have a more specific role in iron mobilization through a specific membrane interaction with iron release sites on the liver cells, in a process which is the reversc

618 E. Baker, A. G. Morton and A. S . Tavill of iron uptake by erythropoietic cells. These alternatives were investigated using the iron chelator desferrioxamine. Desferrioxamine was as effective at mobilizing hepatic iron as transferrin, per unit iron-binding site, and its ability to mobilize liver iron was similarly a function of the prelabelling time (Figs 2 and 7). This suggests that both iron chelators mobilize the same iron pool, possibly by the same mechanism.

In studies on iron efflux from the perfused dog and pig liver, Osaki et a1 (1971) found that caeruloplasmin was essential for iron mobilization, while apotransferrin was almost ineffec- tive. In contrast, in the present study apotransferrin was an efficient mobilizer of iron, in the absence of added caeruloplasmin. There are several possible reasons for this difference. Different perfusion techniques were used in the two studies. In the present work the liver was oxygenated to a physiological level using rat erythrocytes, while Osaki et al (1971) used a buffer perfusate containing no plasma proteins or erythrocytes. Hypoxia is known to increase liver iron mobilization (Mazur et al, 1955). Also, Osaki et a2 (1971) measured total %Fe efflux from reticuloendothelial and parenchymal cells, while in the present study 59Fe release was measured after selective labelling of the parenchymal cells. Hence, differences in the regulation of parenchymal and reticuloendothelial iron mobilization may be involved. Similarly, species differences in the mechanisms regulating liver iron mobilization may also be relevant. While rat caeruloplasmin has a much lower ferroxidase activity than pig or human caeruloplasmin this property is probably not important in its iron mobilization action (Williams et al, 1974). Transferrin gave the same iron release response in media containing commercial albumin or purified albumin, suggesting that any caeruloplasmin contaminant in the albumin preparation was not causing the observed response. Also, human and rat transferrin were equally effective in increasing iron efflux from the rat liver, indicating that any species difference in transferrin was not likely to be important.

Iron efflux into plasma increased when iron containing transferrin was added, but not when iron free transferrin was added. In this series of experiments (Fig 5) there was also an apparent increase in iron mobilization as the transferrin iron content of the plasma perfusate was increased. This can be compared with the net increase in iron uptake by the perfused liver that is observed when the iron saturation of the perfusing medium is increased (Morton & Tavill, 1978). However, it also occurred when the addition of transferrin-bound iron produced no alteration in the unsaturated iron binding capacity. Therefore, it appears that as the ambient mass of circulating transferrin-bound iron is increased so does the turnover between intrahepa- tic iron and the plasma transferrin iron pool increase in absolute terms. It is likely that this increased turnover occurs primarily within the chelatable iron pool, which in turn promotes iron exchange with a more labile component of ferritin iron (Hoy et al, 1974).

It thus seems probable that the iron efflux measured in the present experiments represents the sum of two processes. Firstly, there is iron released by a specific iron mobilization process requiring apotransferrin or another chelator, which is responsible for net efflux. Secondly, there is continuing iron exchange between perfusate tran~ferrin-~~Fe and a pool of hepato- ~yte-~’Fe which reflects the process of iron turnover in the body iron stores and which in normal circumstances is in a steady state between uptake and release. Perturbation of this balance by increasing the plasma transferrin-iron pool (without necessarily changing the plasma unsaturated iron binding capacity) or the plasma transferrin-iron saturation promotes both uptake and release. In the case of increase in saturation there is evidence that there is

Iron Releasefrom Perjiised Rat Liver 619

stimulation of the former more than the latter leading to net deposition of iron. In both these processes the labile iron pool within the hepatocyte may play a vital intermediary role between transport iron and the more stable ferritin iron pool.

ACKNOWLEDGMENTS

Dr A, G. Morton was supported at the Clinical Research Centre by a studentship from the Medical Research Council. Dr E. Baker was supported by the Medical Research Council and the Wellcome Trust. The authors wish to acknowledge valuable discussions of this work with Professor E. R. Huehns. Professor E. €1. Morgan provided useful comments on the manuscript.

REFERENCES

BAKER, E., MORTON, A.G. & TAVILL, A.S. (1975) The effect o f transferrin on iron release from the pcr- fused liver. Proteins oflron Storage and Transport in Biochemistry and Medicine (ed. by R. R. Crichton), pp, 173-180. North Holland Publishing Company, Amsterdam.

BATES, G.W., BILLUPS, C. & SALIMAN, P. (1967) The kinetics and mechanism of iron (111) exchange between chelators and transferrin.Journal ofBiologi- cal Chemistry, 242,2810-2815.

COOK, J.D., HERSHKO, C. & FINCH, C.A. (1973) Stor- age iron kinetics. V. Iron exchange in the rat. British

Journal ofHaematology, 25,6957oh.

isolation of ferritin for the assay of the incorpo- ration of ''C-labelled amino acids. Biochemical Jour- nal, 95,851-858.

DRYSDALE, J.W. & MUNRO, H.N. (1966) Regulation of synthesis and turnover of ferritin in rat liver. Journal of Biological Chemistry, 241,3630-3637.

EAGLE, H. (1955) The minimum vitamin require- ments of the L and HeLa cells in tissue culture, the production of specific vitamin deficiencies and their cure. Journal of Experimental Medicine, 102, 595- 602.

GARDINER, M.E. & MORGAN, E.H. (1974) Trdnsferrin and iron uptake by the liver in thc rat. Australian Jourfial of Experimental Biology and Mcdical Science, 52,723-736.

GORDON, A.H. & LOUIS, L.N. (1963) Preparation and properties of rat transferrin. BiochemicalJournal, 88, 409-414.

HALLBERG, L. & SOLVELL, L. (1960) Absorption of a single dose of iron in man. Acta Medico Scandinavica,

HARRIS, D.C. & AISEN, P. (1975) Iron-donating pro- perties of transferrin. Biochemistry, 14, 262-268.

HERSHKO, C., COOK, J.D. & FINCH, C.A. (1973) Stor- age iron kinetics. 111. Study of desferrioxamine action by selective radioiron labels of reticuloen-

DRYSDI4LE. J.W. & MUNRO, H.N. (1965) Small-scale

168, SUPPI. 358, 19-42.

dothelial and parenchymal cells. Journal of Labora- tory and Clinical Medicine, 81, 876486.

HOY, T.G., HARRISON, P.M. & SHABBIR, M. (1974) Uptake and release of ferritin iron: surface effects and exchange within the crystalline core. Biorhemi- calJournal, 139,605407.

JACOBS, A. (1977) An intracellular transit iron pool. Iron Metabolism, pp. 91-106. Ciba Foundation Symposium 51 (new series).

JANDL, J.H. 81 KATZ, J.H. (1963) The plasma to cell cycle of transferrin. Journal of Clinical Investigation, 42,314-326.

LAURELL, C.B. &MORGAN, E.H. (1964) Iron exchange between transferrin and the placenta in the rat. Acta Physiologica Scandinavica, 62, 271-279.

LINDER, M.C. & MUNRO, H.N. (1972) Assay of tissue ferritin. Analytical Biochemistry, 48,266-278.

DUGARD, J., BOTHWELL, T.H. & CHAHLTON, R.W. (1971) Some factors affecting the release of iron from reticuloendothelial cells. BritishJournal o f H a ~ - matology, 21, 289-303.

MAZUR, A., BAEZ, S . & SHORR, E. (1955) The mechanism ofiron release from ferritin as related to its biological properties.Journal ofBiological Chemis- try, 213, 147-160.

MILSOM, J.P. & BATEY, R . G . (1978) Metabolism of transferrin-hound iron by the liver: a study in t h o . Biochemical Society Transactions, 6,966-968.

MORTON, A.G. & T A v I L L , A.S. (1977) The role ofiron in the regulation of hepatic transferrin synthesis. British Journal of Haematology, 36, 383-394.

MORTON, A.G. & TAVILL, A S . (1978) The control of hepatic iron uptake: Correlation with transferrin synthesis. British Journal of Haematology, 39,

O S A K I , s., JOHNSON, D. & FRIEDEN, E. (1971) The mobilization of iron from perfused mammalian liver by a serum copper enzyme, ferroxidase 1. Journal oJBiological Chemistry, 246, 3018-3023.

LIPSCHKZ, D.A., SIMON, M.O., LYNCH, S.R.,

497-507.

620 E. Baker, A. G. Morton and A. S . Tavill PRICE, E.M. & GIBSON, J.F. (1972) A reinterpretation

of bicarbonate-free ferric transferrin E.P.R. spectra. Biochemical and Biophysical Research Communications, 46,646-651.

SIIMES, M.A. & DALLMAN, P.R. (1974) New kinetic role for serum ferritin in iron metabolism. British

Journal ofHaematology, 28, 7-18. TSAO, M.Y. & BELILES, R.P. (1973) A modified micro

method for determination of iron in tissue. Ameri- can Journal of Clinical Pathology, 59, 1 6 1 6 6 .

UNGER, A. & HERSHKO, C. (1974) Hepatocellular uptake of ferritin in the rat. British Journal of Haema- tology, 28, 169-179.

VAN WYK, C.P., LINDER-HOROWITZ, M. & MUNRO, H.N. (1971) Effect of iron loading on non-heme

iron compounds in different liver cell populations. Journal ofBiologicaZ Chemistry, 246,1025-1031.

WILLIAMS, D.M., LEE, G.R. & CARTWRIGHT, G.E. (1974) Ferroxidase activity of rat ceruloplasmin. American Journal OfPhysiology, 227, 1094-1097.

WORWOOD, M. (1979) Serum ferritin. CRC Critical Reviews in Clinical Laboratory Sciences, 10, 171-204.

YOUNG, D.S. & HICKS, J.M. (1965) Method for the automatic determination of serum iron. Journal of Clinical Pathology, 18,98-102.

ZUYDERHOUUT, F.M.J., LIMHORST, C. & HENGEVELD, P. (1978) On the iron content of human serum ferritin, especially in acute viral hepatitis and iron overload. CZinica Chimica Acta, 90, 93-99.