Uptake of iron from transferrin by isolated hepatocytes

Isolated rat hepatocytes containing 0.56-1.79 micrograms iron/10(6) cells and with an intracellular ATP concentration of 3-4 mM, accumulate iron from transferrin linearly with time for at least 3 h. At 37 degrees C the rate of uptake amounts to 0.3-0.7 pmol/mg cell protein per min. The uptake reaches a saturation level of 21-40 pmol/mg cell protein per h at 2.2 microM iron. At 5 degrees C the uptake does not increase over the time of incubation. Uptake of iron, but not binding of transferrin is increased 4-5-fold at oxygen concentrations 10-20 microM. At oxygen concentrations beyond these limits iron uptake is decreased. Iron taken up at low oxygen concentrations can be chelated by bathophenanthroline and bathophenanthroline disulphonate , but only if the chelators are present during the uptake experiments. The results suggest that iron uptake from transferrin by hepatocytes in suspension involves reductive removal of iron.     

Albumin inhibition of transferrin low-affinity binding to K562 cells

Several reports have suggested that variations of albumin concentration in the incubation medium can modulate the magnitude of transferrin binding to the cells. We have investigated this problem further using K562 cells. In the absence of human serum albumin, transferrin binding demonstrated a non-saturable curve which, upon Scatchard analysis, showed two components with high and low affinities. In the presence of 0.5% human serum albumin, the low-affinity but not the high-affinity component was totally inhibited and, thus, the binding showed a saturation plateau at transferrin concentration of 6 micrograms/ml. Increasing concentrations of human serum albumin in the incubation medium led to progressive inhibition of transferrin binding, reaching a plateau at 0.2% human serum albumin. At this concentration transferrin binding was about 12 ng/10(6) cells, corresponding to the saturation plateau for high-affinity binding. Low-affinity transferrin binding in the absence of human serum albumin could readily be displaced by subsequent addition of albumin. Similar inhibition was obtained by another serum protein, ceruloplasmin, suggesting that this inhibition is not unique to albumin and may be a common property of all proteins. Incubation at 37 degrees C with 59Fe-labeled transferrin indicated that all iron uptake occurs through high-affinity binding. We conclude that the reported variations in magnitude of transferrin binding by the cell due to variations in albumin concentration are the result of inhibition of low-affinity binding of transferrin by albumin.     

Albumin prevents nonspecific transferrin binding and iron uptake by isolated hepatocytes

Bovine serum albumin inhibits binding of transferrin by hepatocytes in suspension by 60-70%. Iron uptake is inhibited by less than 20%. A Scatchard analysis of the transferrin-binding data reveals a biphasic plot in the absence of bovine serum albumin, but a monophasic plot in the presence of bovine serum albumin. Bovine serum albumin inhibits low-affinity binding of transferrin (125000 molecules/cell), but has no effect on high-affinity binding (38000 molecules/cell). In pronase-treated cells, transferrin binding is reduced by 40%, and when bovine serum albumin is added, the binding is reduced by a further 40%. Corresponding figures for iron uptake are 70 and 10%, respectively. The results are strong evidence that the major part of iron uptake by hepatocytes occurs from transferrin bound to the plasma membrane transferrin receptor.     

Chronic iron overload inhibits protein secretion by adult rat hepatocytes maintained in long-term primary culture

Short-term pure cultures and long-term cocultures of adult rat hepatocytes with rat liver epithelial cells, presumably derived from primitive biliary cells, were used to define in vitro models of iron overloaded hepatocytes in order to understand the molecular mechanism responsible for liver damage occurring in patients with hemochromatosis. In vitro iron overload was obtained by daily addition of ferric nitrilotriacetate to the culture medium. A concentration of 20 microM ferric salt induced hepatocyte iron overload with minimal cytotoxicity as evaluated by cell viability, morphological changes of treated cells and cytosolic enzyme leakage into the culture medium. The effects of iron overload on protein biosynthesis and secretion were studied in both short-term pure cultures and long-term cocultures of hepatocytes. The amounts of intracellular and newly synthesized proteins were never modified by the iron treatment. Furthermore, neither the relative amounts of transferrin and albumin mRNAs nor their translational products were altered by iron overload. Moreover, no change in the transferrin isomeric forms were observed in treated cells. In contrast, a prolonged exposure of cocultured hepatocytes to 20 microM ferric salt led to a significant decrease in the amount of proteins secreted in the medium. This decrease included the two major secreted proteins, namely albumin and transferrin, and probably all other secreted proteins. These results demonstrate that iron loading alters neither the total nor the liver specific protein synthesis activity of cultured hepatocytes. They suggest that chronic overload may impede the protein secretion process.     

Transferrin-bound and transferrin free iron uptake by cultured rat astrocytes.

Previously we had demonstrated the presence of transferrin receptor (TfR) on the plasma membrane of cultured rat cortical astrocytes. In this study, we investigated the roles of TfR in transferrin-bound iron (Tf-Fe) as well as transferrin-free iron (Fe II) uptake by the cells. The cultured rat astrocytes were incubated with 1 microM of double-labelled transferrin (125I-Tf-59Fe) in serum- free DMEM F12 medium or 59Fe II in isotonic sucrose solution at 37 degrees C or 4 degrees C for varying times. The cellular Tf-Fe, Tf and Fe II uptake was analyzed by measuring the intracellular radioactivity with gamma counter. The result showed that Tf-Fe uptake kept increasing in a linear manner at least in the first 30-min. In contrast to Tf-Fe uptake, the internalization of Tf into the cells was rapid initially but then slowed to a plateau level after 10 min. of incubation. The addition of either NH4Cl or CH3NH2, the blockers of Tf-Fe uptake via inhibiting iron release from Tf within endosomes, decreased the cellular Tf-Fe uptake but had no significant effect on Tf uptake. Pre-treated cells with trypsin inhibited significantly the cellular uptake of Tf-Fe as well as Tf. These findings suggested that Tf-Fe transport across the membrane of astrocytes is mediated by Tf-TfR endocytosis. The results of transferrin-free iron uptake indicated that the cultured rat cortical astrocytes had the capacity to acquire Fe II. The highest uptake of Fe II occurred at pH 6.5. The Fe II uptake was time and temperature dependent, iron concentration saturable, inhibited by several divalent metal ions, such as Co2+, Zn2+, Mn2+ and Ni2+ and not significantly affected by phenylarsine oxide treatment. These characteristics of Fe II uptake by the cultured astrocytes suggested that Fe II uptake is not mediated by TfR and implied that a carrier-mediated iron transport system might be present on the membrane of the cultured cells.     

Heme inhibits transferrin endocytosis in immature erythroid cells

The inhibitory effect of heme on iron uptake from transferrin by rat and rabbit reticulocytes and erythroid cells from the fetal rat liver was studied in vitro. Addition of hemin was shown to cause a decrease in the rate of transferrin endocytosis, the degree of inhibition being proportional to the reduction in iron uptake. The heme synthesis inhibitors, isoniazid and succinylacetone, stimulated the rate of transferrin endocytosis by 15-30% and caused a proportional increase in the rate of iron uptake, possibly by reducing the intracellular free heme concentration. It is concluded from these results that heme affects iron uptake by influencing the rate of transferrin endocytosis and recycling.     

Hepatocytes and reticulocytes have different mechanisms for the uptake of iron from transferrin

The uptake of iron from transferrin by isolated rat hepatocytes and rat reticulocytes has been compared. The results show the following. 1) Reticulocytes and hepatocytes express plasma membrane NADH:ferricyanide oxidoreductase activity. The activity, expressed per 10(6) cells, is approximately 60-fold higher in the hepatocyte than in the reticulocyte. 2) Hepatocyte plasma membrane NADH:ferricyanide oxidoreductase activity and uptake of iron from transferrin are stimulated by low oxygen concentration and inhibited by iodoacetate. In reticulocytes, similar changes are seen in NADH:ferricyanide oxidoreductase activity, but not on iron uptake. 3) Ferricyanide inhibits the uptake of iron from transferrin by hepatocytes, but has no effect on iron uptake by reticulocytes. 4) Perturbants of endocytosis and endosomal acidification have no inhibitory effect on hepatocyte iron uptake, but inhibit reticulocyte iron uptake. 5) Hydrophilic iron chelators effectively inhibit hepatocyte iron uptake, but have no effect on reticulocyte iron uptake. Hydrophobic iron chelators generally inhibit both hepatocyte and reticulocyte iron uptake. 6) Divalent metal cations with ionic radii similar to or less than the ferrous iron ion are effective inhibitors of hepatocyte iron uptake with no effect on reticulocyte iron uptake. The results are compatible with hepatocyte uptake of iron from transferrin by a reductive process at the cell surface and reticulocyte iron uptake by receptor-mediated endocytosis.     

Measurement of saturable and non-saturable components of anandamide uptake into P19 embryonic carcinoma cells in the presence of fatty acid-free bovine serum albumin

There is considerable controversy at present concerning the mechanisms responsible for the cellular uptake of anandamide. One particular issue concerns whether fatty acid-free bovine serum albumin should be used in the assays, it having been argued that such a presence effectively prevents the specific uptake of anandamide. In the present study, it has been demonstrated that in the presence of a low (0.1%, w/v) concentration of fatty acid-free bovine serum albumin, a temperature-dependent and saturable (K(m) approximately 1 microM) uptake of anandamide into P19 embryonic carcinoma cells can be demonstrated using an incubation time of 4 min. Under these conditions, the uptake of anandamide at 4 degrees C is low at a substrate concentration of 100 nM. The uptake at 37 degrees C was not significantly reduced following treatment of the cells with either methyl-beta-cyclodextrin (50 microM) or mevinolin (1 microM), but was reduced by the FAAH inhibitor URB597 (1 microM) and inhibited by the transport inhibitor cum FAAH substrate AM404 with an IC(50) value of 12 microM. When a 45 s incubation time was used, the uptake of anandamide was not saturable at 37 degrees C over the concentration range tested (0.1-1 microM). Analysis of the data at 37 degrees C obtained with 45 s, 4 min and 15 min incubation times revealed a very rapid (i.e. complete by 45 s) non-saturable component followed by a slower saturable (K(m) approximately 1 microM) component of the uptake. It is concluded that the presence of a low concentration of fatty acid-free bovine serum albumin at a suitable concentration reduces non-specific binding (and release) of anandamide to cell culture wells, greatly reduces the cellular accumulation seen at 4 degrees C, and allows the visualisation of both non-saturable and saturable components of the uptake to be seen at 37 degrees C.     

Transferrin binding and iron uptake in mouse hepatocytes

Methods were developed for obtaining highly viable mouse hepatocytes in single cell suspension and for maintaining the hepatocytes in adherent static culture. The characteristics of transferrin binding and iron uptake into these hepatocytes was investigated. (1) After attachment to culture dishes for 18–24 h hepatocytes displayed an accelerating rate of iron uptake with time. Immediately after isolation mouse hepatocytes in suspension exhibited a linear iron uptake rate of 1.14·10⁵molecules/cell per min in 5 μM transferrin. Iron uptake also increased with increasing transferrin concentration both in suspension and adherent culture. Pinocytosis measured in isolated hepatocytes could account only for 10–20% of the total iron uptake. Iron uptake was completely inhibited at 4°C. (2) A transferrin binding component which saturated at 0.5 μM diferric transferrin was detected. The number of specific, saturable diferric transferrin binding sites on mouse hepatocytes was 4.4·10⁴±1.9·10⁴ for cells in suspension and 6.6·10⁴±2.3·10⁴ for adherent cultured cells. The apparent association constants were 1.23·10⁷ 1·mol^?1 and 3.4·10⁶ 1·mol^?1 for suspension and cultured cells respectively. (3) Mouse hepatocytes also displayed a large component of non-saturable transferrin binding sites. This binding increased linearly with transferrin concentration and appeared to contribute to iron uptake in mouse hepatocytes. Assuming that only saturable transferrin binding sites donate iron, the rate of iron uptake is about 2.5 molecules iron/receptor per min at 5 μM transferrin in both suspension and adherent cells and increases to 4 molecules iron/receptor per min at 10 μM transferrin in adherent cultured cells. These rates are considerably greater than the 0.5 molcules/receptor per min observed at 0.5 μM transferrin, the concentration at which the specific transferrin binding sites are fully occupied. The data suggest that either the non-saturable binding component donates some iron or that this component stimulates the saturable component to increase the rate of iron uptake. (4) During incubations at 4°C the majority of the transferrin bound to both saturable and nonsaturable binding sites lost one or more iron atoms. Incubations including 2 mM α,α′-dipyridyl (an Fe¹¹ chelator) decreased the cell associated ⁵⁹Fe at both 4 and 37°C while completely inhibiting iron uptake within 2–3 min of exposure at 37°C. These observations suggest that most if not all iron is loosened from transferrin upon interaction of transferrin with the hepatocyte membrane. There is also greater sensitivity of ⁵⁹Fe uptake compared to transferrin binding to pronase digestion, suggesting that an iron acceptor moiety on the cell surface is available to proteolysis.     

Evidence for the functional equivalence of the iron-binding sites of rat transferrin

The role of the two iron-binding sites of rat transferrin in the exchange of iron with cells has been assessed using urea polyacrylamide gel electrophoresis to separate and quantitate the four possible molecular species of transferrin generated during the incubation of ¹²⁵I-labelled transferrin with rat reticulocytes and hepatocytes. Addition of diferric transferrin to reticulocytes led directly to the appearance of apotransferrin together with small and comparable amounts of the two monoferric transferrins. After 2 h 44.8% of the iron had been removed by the cells, and of the iron-depleted transferrin 71.8% was apotransferrin, the remainder being monoferric transferrin, 16.1% with N-terminal iron and 12.1% with C-terminal iron. A similar pattern emerged with hepatocytes, but the rate of iron removal was slower and the proportion of apotransferrin generated was lower. After 4 h 10.9% of the iron had been removed from the transferrin and the distribution of the iron-depleted protein was: apotransferrin 26.9% and monoferric (N-terminal) 39.2%, (C-terminal) 33.9%. The appearance of apotransferrin during each incubation and the generation of both monoferric transferrins suggest that both cell types are able to remove iron from differic transferrin in pairwise fashion and that they do not appreciably distinguish between the two iron-binding sites of the protein. Release of iron from hepatocytes to apotransferrin lead to the appearance of both monferric species and then to increasing amounts of diferric transferrin. The process of iron release did not seem to distinguish between the vacant iron-binding sites of transferrin.     

Intracellular processing of transferrin and iron by isolated rat hepatocytes.

Transferrin bound by isolated rat hepatocytes is rapidly endocytosed and enters a compartment of low density. Little was found associated with the lysosomes, even though the protein was subsequently lost from the cells. Iron entering the cells on transferrin was subsequently found in a number of intracellular components: transferrin, haem, ferritin and a residual fraction. After 2 h incubation with 59Fe-transferrin almost 70% of the iron was in ferritin, and this proportion increased to 80% during a 'chase' experiment. Residual iron, because of its rapid increase at the start of the incubation and its decline during the 'chase', probably represents an intracellular transit pool, which at steady state was present at 23 pg/10(6) cells.     

Uptake of ferritin and iron bound to ferritin by rat hepatocytes: modulation by apotransferrin, iron chelators and chloroquine

Rat liver ferritin is an effective donor of iron to rat hepatocytes. Uptake of iron from ferritin by the cells is partially inhibited by including apotransferrin in the culture medium, but not by inclusion of diferric transferrin. This inhibition is dependent on the concentration of apotransferrin, with a 30% depression in iron incorporation in the cells detected at apotransferrin concentrations above 40 micrograms/ml. However, apotransferrin does not interfere with uptake of 125I-labeled ferritin, suggesting that apotransferrin decreases retention of iron taken up from ferritin by hepatocytes by sequestering a portion of released iron before it has entered the metabolic pathway of the cells. The iron chelators desferrioxamine (100 microM), citrate (10 mM) and diethylenetriaminepentaacetate (100 microM) reduce iron uptake by the cells by 35, 25 and 8%, respectively. In contrast, 1 mM ascorbate increases iron accumulation by 20%. At a subtoxic concentration of 100 microM, chloroquine depresses ferritin and iron uptake by hepatocytes by more than 50% after 3 h incubation. Chloroquine presumably acts by retarding lysosomal degradation of ferritin and recycling of ferritin receptors.     

Growth-stimulating effect of transferrin on a hybridoma cell line: Relation to transferrin iron-transporting function

The relation of the growth-stimulating capacity of transferrin to its iron-transporting function was investigated in mouse hybridoma PLV-01 cells cultivated in a chemically defined medium. The cells were precultivated in protein-free medium supplemented either with ferric citrate (cells with a high intracellular iron level) or with iron-saturated transferrin (cells with a low intracellular iron level). Iron uptake was monitored after the application of 59Fe-labeled ferric citrate or pig transferrin. Cultivation of the cells at the optimum growth-stimulating concentration (500 microM) of ferric citrate resulted in an intracellular iron level about 100-fold higher than that of cells cultivated at the optimum transferrin concentration (5 micrograms/ml). Replacement of pig transferrin with bovine transferrin resulted in similar intracellular iron levels, but the growth-stimulating effect of bovine transferrin was more than one order of magnitude lower. Cells with a high intracellular iron level grew equally well when cultivated with iron-saturated transferrin or with apotransferrin + deferoxamine (2 micrograms/ml). On the other hand, cells with a low intracellular iron level required iron-saturated transferrin for further growth and apotransferrin + deferoxamine was ineffective. The results suggest that transferrin can act as a cell growth factor only in the iron-saturated form. However, several findings of this work indicate that supplying cells with iron cannot be accepted as the full explanation of the transferrin growth-stimulating effect.     

Iron metabolism pathways in the rat hepatocyte

A small to moderate inhibitory effect of iron uptake by isolated rat hepatocytes in short-term studies was seen with oxidative phosphorylation and electron transport inhibitors, and no inhibition by agents affecting pinocytosis. Intracellular transferrin was able to donate iron to the small-molecular weight iron pool, and the latter was able to transfer, by a process not requiring energy or movement of serum transferrin, iron to ferritin. Serum transferrin was not able to lose iron to any cytosol components. Reducing agents were not able to abstract iron from rat serum transferrin to any great extent. It is concluded that iron is taken up by the rat hepatocyte from serum transferrin by a process not requiring energy or movement of serum transferrin into the cell interior; and that intracellular transferrin is involved in acquiring iron from serum transferrin at the cell surface, with iron then being transferred to the small-molecular weight iron pool and hence to ferritin. It is also proposed that intracellular transferrins may have the general function of interacting with serum transferrin at cell surfaces.     

Release of iron from diferric transferrin in the presence of rat liver plasma membranes: no evidence of a plasma membrane diferric transferrin reductase

The transfer of iron from diferric transferrin to bathophenanthroline disulfonate was measured under varying conditions by spectrophotometry and EPR spectroscopy. Intact rat hepatocytes efficiently mediated the transfer of iron from human diferric transferrin to bathophenanthroline disulfonate. Isolated rat liver plasma membranes, in contrast, failed to facilitate the reaction at pH 7.4 in the presence of NADH, although the membranes were able to reduce ferricyanide and to oxidize NADH. Oxidation of NADH was stimulated by diferric transferrin. However, ferricyanide reductase and transferrin-stimulated NADH oxidase activities were apparently not linked to release of iron from transferrin. Our results, together with theoretical considerations, show that the ability (or inability) of intact cells or isolated plasma membranes to facilitate the transfer of iron from transferrin to strong diferric iron chelators does not allow interferences about the existence of an iron reduction step as part of the process of cellular uptake of iron from transferrin.     

Effect of nicotine on transferrin binding and iron uptake by cultured rat placenta

The effect of nicotine on transferrin and iron transport in placental cells has been studied. Nicotine inhibits iron uptake but has little effect on the steady-state levels of transferrin. The effect is temperature and concentration dependent and is not reversible. At a concentration of 15 mM nicotine inhibited transferrin endocytosis by 40%, while iron uptake was decreased by nearly 60%. Nicotine exerted a similar effect on reticulocytes, but other amines, either tertiary or quaternary, had little or no effect on either iron uptake or steady-state intracellular transferrin levels. The results suggest that nicotine acts by blocking uptake, probably by acting as a weak base inhibiting iron release from transferrin, and inhibiting exocytosis with a resultant block of endocytosis. The concentrations required to exert an effect are too high to implicate inhibition of iron transport in the effects of smoking on pregnancy.     

Uptake and subcellular processing of 59Fe-125I-labelled transferrin by rat liver.

The uptake of transferrin and iron by the rat liver was studied after intravenous injection or perfusion in vitro with diferric rat transferrin labelled with 125I and 59Fe. It was shown by subcellular fractionation on sucrose density gradients that 125I-transferrin was predominantly associated with a low-density membrane fraction, of similar density to the Golgi-membrane marker galactosyltransferase. Electron-microscope autoradiography demonstrated that most of the 125I-transferrin was located in hepatocytes. The 59Fe had a bimodal distribution, with a larger peak at a similar low density to that of labelled transferrin and a smaller peak at higher density coincident with the mitochondrial enzyme succinate dehydrogenase. Approx. 50% of the 59Fe in the low-density peak was precipitated with anti-(rat ferritin) serum. Uptake of transferrin into the low-density fraction was rapid, reaching a maximal level after 5-10 min. When livers were perfused with various concentrations of transferrin the total uptakes of both iron and transferrin and incorporation into their subcellular fractions were curvilinear, increasing with transferrin concentrations up to at least 10 microM. Analysis of the transferrin-uptake data indicated the presence of specific transferrin receptors with an association constant of approx. 5 X 10(6) M-1, with some non-specific binding. Neither rat nor bovine serum albumin was taken up into the low-density fractions of the liver. Chase experiments with the perfused liver showed that most of the 125I-transferrin was rapidly released from the liver, predominantly in an undegraded form, as indicated by precipitation with trichloroacetic acid. Approx. 40% of the 59Fe was also released. It is concluded that the uptake of transferrin-bound iron by the liver of the rat results from endocytosis by hepatocytes of the iron-transferrin complex into low-density vesicles followed by release of iron from the transferrin and recycling of the transferrin to the extracellular medium. The iron is rapidly incorporated into mitochondria and cytosolic ferritin.     

The effect of lysosomotrophic bases and inhibitors of transglutaminase on iron uptake by immature erythroid cells

The mechanism by which weak bases block iron uptake by immature erythroid cells was investigated using rabbit and rat reticulocytes and erythroblasts from the fetal rat liver. A large variety of bases was found to inhibit iron uptake but to have a much smaller or no effect on transferrin uptake by the cells. Quinacrine and chloroquine were active at the lowest concentrations. Dansylcadaverine, an inhibitor of transglutaminase, was also active at low concentration. However, the results do not indicate a role for transglutaminase in the iron uptake process. Instead they show that the major effect of the bases is to inhibit iron release from transferrin molecules on or within the cells. The possible mechanism of this effect was investigated by measurement of intracellular ATP levels, intracellular pH and by morphological studies utilizing fluorescent and electron microscopy. The bases caused little change in ATP levels, but elevated intracellular pH, probably due to accumulation within intracellular vesicles, which were shown to accumulate fluorescent weak bases, to swell under the action of the bases and to be the site of intracellular localization of transferrin. It is concluded that the bases tested in this work inhibit iron release from transferrin in intracellular vesicles by increasing their pH rather than by blocking transglutaminase and thereby restricting endocytosis. Reduction of transferrin uptake by the cells when it occurs is probably due to inhibition of recycling of transferrin receptors to the outer cell membrane.     

Regulators of fetal liver differentiation in vitro

Seventeen-day-old fetal rat hepatocytes were employed to examine factors required to promote differentiation in vitro. In the absence of effectors, primary fetal hepatocytes dedifferentiated, as characterized by the rapid decline in synthesis of fetal alpha-fetoprotein (AFP), albumin, and transferrin. On the other hand, cells maintained in the presence of glucocorticoid hormone produced high levels of albumin and transferrin. Glucocorticoid could not prevent the decline in fetal AFP synthesis, but induced synthesis of the 65K variant AFP--the major AFP species produced by adult rat liver. Fetal hepatocytes maintained in the presence of 8-bromo-cAMP (8-BrcAMP), or methyl isobutyl xanthine (MIX), an agent that increases intracellular cAMP levels, synthesized high levels of fetal AFP and albumin but reduced levels of transferrin. Both glucocorticoid and 8-BrcAMP or MIX induced expression of adult liver-specific genes such as tyrosine aminotransferase (TAT) and phosphoenolpyruvate carboxykinase (PEPCK), suggesting that these fetal hepatocytes have matured. Cells maintained in the presence of glucocorticoid hormone and MIX (or 8-BrcAMP) contained more albumin, TAT, and PEPCK mRNAs and synthesized increased amounts of the 65K variant AFP than those with either agent alone. However, the glucocorticoid/MIX cells produced intermediate levels of the fetal AFP and transferrin. Our data indicate that both glucocorticoid hormone and cAMP are necessary for optimal differentiation of fetal hepatocytes in vitro.