Astrocytes are central to iron and ascorbate homoeostasis within the brain. Although NTBI (non-transferrin-bound iron) may be a major form of iron imported by astrocytes in vivo, the mechanisms responsible remain unclear. The present study examines NTBI uptake by cultured astrocytes and the involvement of ascorbate and DMT1 (divalent metal transporter 1). We demonstrate that iron accumulation by ascorbate-deficient astrocytes is insensitive to both membrane-impermeant Fe(II) chelators and to the addition of the ferroxidase caeruloplasmin. However, when astrocytes are ascorbate-replete, as occurs in vivo, their rate of iron accumulation is doubled. The acquisition of this additional iron depends on effluxed ascorbate and can be blocked by the DMT1 inhibitor ferristatin/NSC306711. Furthermore, the calcein-accessible component of intracellular labile iron, which appears during iron uptake, appears to consist of only Fe(III) in ascorbate-deficient astrocytes, whereas that of ascorbate-replete astrocytes comprises both valencies. Our data suggest that an Fe(III)-uptake pathway predominates when astrocytes are ascorbate-deficient, but that in ascorbate-replete astrocytes, at least half of the accumulated iron is initially reduced by effluxed ascorbate and then imported by DMT1. These results suggest that ascorbate is intimately involved in iron accumulation by astrocytes, and is thus an important contributor to iron homoeostasis in the mammalian brain.
Although iron is essential for normal brain function, an excess of labile iron can cause oxidative damage to brain tissue [1,2]. Therefore, iron uptake into the brain is usually tightly regulated. Under normal conditions, many mammalian cells (e.g. erythroid precursors  and neurons ) acquire most of their iron from the serum iron-carrier protein Tf (transferrin). In addition to TBI (Tf-bound iron), however, most cells are capable of the high-capacity uptake of NTBI (non-TBI). NTBI is thought to be iron that is bound to a dynamic pool of low-molecular-mass ligands such as citrate . In cells from tissues other than the brain (e.g. liver), the uptake of NTBI is usually only considered to be relevant under conditions of iron overload, when the amount of iron in the serum exceeds the iron-binding capacity of Tf .
Brain cells are unusual since under normal conditions they appear to be adapted either for the uptake of TBI (e.g. neurones) or NTBI (e.g. astrocytes, oligodendrocytes and microglia) [1,2,5]. The majority of iron enters the brain through the uptake of plasma TBI by brain capillary endothelial cells, followed by the release of NTBI into the interstitial spaces of the brain . Much of this NTBI is probably complexed by molecules such as citrate, and is then taken up by the endfeet of astrocytes that are in close association with brain capillary endothelial cells . The mechanisms responsible for the uptake of NTBI by astrocytes are not known with certainty . DMT1 (divalent metal transporter 1) is highly expressed in astrocytic endfeet both in culture [6,7] and in vivo . Moreover, DMT1 levels are acutely regulated by cellular iron status in primary astrocyte cultures . These observations are consistent with the likelihood that DMT1 provides a major route for the uptake of Fe(II) into astrocytes. Direct evidence for this hypothesis has been lacking, however, and it was recently suggested that although cultured astrocytes are capable of Fe(II) uptake, a putative Fe(III)-selective uptake pathway predominates in the presence of 100 μM of FAC (ferric ammonium citrate) .
The DMT1-dependent uptake of NTBI by cells typically requires Fe(III) to be reduced to Fe(II) . As with most mammalian cells, the mechanism of NTBI ferrireduction by astrocytes is unresolved , but it probably involves transplasma membrane electron transport to facilitate the reduction of Fe(III) prior to the uptake of Fe(II) [1,11]. Whereas most models of cellular ferrireduction incorporate membrane-bound ferrireductases , we have recently implicated a non-enzymatic ferrireduction system involving the export of ascorbate [11–13]. In particular, we have shown that the reduction and uptake of NTBI by human erythroleukaemia (K562) cells is enhanced by ascorbate efflux . According to this model, ascorbate that is released from cells markedly stimulates the reduction of extracellular Fe(III) and the uptake of the resulting Fe(II) .
Ascorbate is a two-electron donor that can be oxidized to DHA (dehydroascorbate), the latter of which can then be imported into cells via facilitative glucose transporters  and reduced to ascorbate by both glutathione-dependent and -independent mechanisms . Within the brain, ascorbate is an essential enzyme cofactor and a neuromodulator, and is present at concentrations of 200–400 μM in extracellular fluid, and at millimolar concentrations within cells . Astrocytes are able to rapidly accumulate ascorbate from extracellular DHA, and then release much of this ascorbate into the extracellular space [11,14]. Astrocytes are considered to play an important role in the brain by regenerating ascorbate from DHA .
In the present study we have examined the mechanisms of NTBI uptake in cultured astrocytes. We provide evidence in support of two independent mechanisms of NTBI uptake: the first contributes to at least 50% of the iron accumulated in ascorbate-replete astrocytes and is apparently Fe(II)-selective, whereas the second predominates in ascorbate-deficient astrocytes and appears to be Fe(III)-selective. As ascorbate is abundant within brain tissue, ascorbate-replete astrocytes may represent the more physiologically relevant scenario. Additionally, our results show that ascorbate-replete astrocytes can release ascorbate from the cytoplasm into the extracellular space, where it reacts directly with ferric citrate to form Fe(II) ions that can then be imported. Moreover, using the DMT1 inhibitor, ferristatin/NSC306711 , we provide the first pharmacological evidence of a role for DMT1 in facilitating the import of Fe(II) into cultured astrocytes. On the basis of our results, ferristatin-sensitive iron accumulation from Fe(II) appears to be active at pH 7.2, which is consistent with the determined pH of the extracellular fluid of the brain . These findings extend our understanding of how astrocytes import iron following its separation from Tf by brain capillary endothelial cells, and also provide insights into the role played by ascorbate in brain iron homoeostasis.
Unless otherwise stated, all chemicals were obtained from Sigma–Aldrich or Merck/Calbiochem. 55Fe radionuclide was purchased in 1 mCi lots as 55Fe-labelled FeCl3 in 0.5 M HCl from PerkinElmer. Calcein-AM (calcein acetoxymethyl ester; Calbiochem) was dissolved to 1 mM in anhydrous DMSO, and stored frozen in aliquots at −20°C. The novel DMT1 inhibitor, ‘NSC compound # 306711’  (referred to as ‘ferristatin’ ) was obtained from the NIH National Cancer Institute Diversity Set library. Ferristatin was dissolved to 7.5 mM in anhydrous DMSO and stored frozen in aliquots at −20°C. The following compounds used in the present study were also dissolved in DMSO: the copper chelator neocuproine (50 mM); the Fe(II) chelator BIP (2,2′-bipyridyl; 20 mM); and the Fe(III) chelator PIH (pyridoxal isonicotinoyl hydrazone; 20 mM; obtained as a gift from Professor Des Richardson, Department of Pathology, University of Sydney, Sydney, NSW, Australia ). DTPA (diethylenetriaminepenta-acetic acid) was dissolved as a stock solution in 0.5 M HCl. Ferric citrate (1:5 molar ratio, iron/citrate) was always made as a stock solution no more than 20 min prior to use by combining one volume of 100 mM FeCl3 dissolved in 1% HCl with ten volumes (i.e. a five-fold molar excess) of 50 mM sodium citrate solution at pH 7.0. This solution was then diluted to the appropriate stock concentration (10-fold higher than the final concentration) with the appropriate incubation medium. The reported molarity of the ferric citrate used in the present study is based on the iron concentration.
Spectrophotometric microplate assays were performed on a Benchmark™ Plus microplate spectrophotometer (BioRad) using Nunc 96-well flat-bottomed transparent plates. Orbital mixing of cell suspensions was performed with a thermoregulated microplate incubator/shaker (Nanjing Foinoe Equipment) set to 37°C.
Astrocyte-rich primary cultures were prepared from the brains of newborn Wistar rats as described previously  and as approved by the Monash University School of Psychology, Psychiatry and Psychological Medicine Animal Ethics Committee. Viable cells were cultured in 24-well culture plates (Greiner Bio-One) as described previously . Immediately prior to use, wells in each plate were washed three times with 1.5–2.0 ml of HBS (Hepes-buffered saline; 134 mM NaCl, 5.2 mM KCl, 1.8 mM CaCl2, 0.8 mM MgSO4 and 20 mM Hepes-Na+, pH 7.2 at 37°C) that had been pre-warmed to 37°C. Where indicated the pH of the HBS solution was adjusted to 6.8 at 37°C.
Ascorbate-loading of cells
Similar to many cell types , astrocytes rapidly become ascorbate deficient when put into culture [14,21,23]. Ascorbate-deficient astrocytes were made ascorbate replete (i.e. loaded with ascorbate) by exposing the cells to freshly prepared solutions of L-DHA dimer for 30 min at 37°C. The reported concentrations of DHA (typically 400 μM) are for the hydrated monomeric species that is formed upon hydrolysis of the crystalline DHA dimer in aqueous solution (i.e. twice the concentration of DHA dimer) . The exposure of astrocytes to DHA was performed in the absence of glucose in order to avoid competitive inhibition of DHA uptake. Following DHA exposure, the extracellular medium was thoroughly removed by aspiration, followed by three successive washes with ice-cold HBS containing 5 mM D-glucose per well, in order to stop DHA uptake and to minimize the efflux of ascorbate from cells. Cells were always incubated with pre-warmed (i.e. 37°C) HBS for 10 min prior to initiation of subsequent assays.
Determination of intracellular ascorbate
55Fe accumulation assays
55Fe accumulation assays were modified from a method described previously . Astrocytes were incubated with 3 μM 55Fe-labelled ferric citrate containing up to 250 nCi/well. The membrane-permeant copper-chelator neocuproine (50 μM) was used during iron-accumulation experiments involving ferristatin in order to chelate labile copper-ions that are present in the ferristatin molecule . For consistency, neocuproine was included in all iron-accumulation assays. Neocuproine did not affect iron accumulation by control cells (results not shown). Following addition of 55Fe-labelled ferric citrate, plates were incubated at 37°C with orbital mixing at 320 rev./min for 60 min. Assays were terminated by the removal of the overlying solution and immediate addition of 1 ml of ice-cold ‘stop solution’ [1 mM DTPA in MBS (Mops-buffered saline, 137 mM NaCl, 2.7 mM KCl and 15 mM Mops-Na+) at a final pH of 6.5]. Cells were washed four times with 1 ml of ice-cold stop solution . Following the final wash, 0.4 ml of 2% (w/v) SDS was added to lyse the cells. Complete cell lysis was achieved by orbital mixing at 37°C at 900 rev./min for 15 min. Cell homogenates were then mixed with 4 ml of liquid scintillation fluid (Optiphase ‘HIsafe’ 2; PerkinElmer) and added to 6 ml plastic scintillation vials. Radioactivity was determined on 3H settings with quench correction in a Wallac 1409 liquid scintillation counter. Iron accumulation was then determined from the measured cell-associated radioactivity and the known specific activity (i.e. pCi/pmol Fe) of the stock solutions of 55Fe-labelled ferric citrate.
Calcein-loading and intracellular calcein-fluorescence-quenching assays
Iron-dependent intracellular calcein fluorescence quenching assays were performed according to a modification of a previous method . Briefly, astrocytes were loaded with the iron-sensitive fluorophore calcein by pre-exposing the cells to 150 nM calcein-AM in HBS for 15 min at 37°C. Cells were loaded with calcein immediately after ascorbate loading, and were subsequently washed three times with ice-cold HBS. The cleavage of intracellular calcein-AM by intracellular esterases is rapid and the resulting free calcein moiety is membrane impermeant provided that membrane integrity is maintained . Calcein is an iron chelator, and when present intracellularly it can be used to detect the influx of iron into the intracellular LIP (labile iron pool) . The LIP is thought to represent a transient state of intracellular iron prior to its entry into ferritin or the mitochondrial iron processing pathways . The binding of imported iron either as Fe(II) or Fe(III) to calcein causes a decrease (or ‘quenching’) of the compound's fluorescence . The specificity of this quenching for iron was determined by reversing the quenched calcein fluorescence (i.e. ‘dequenching’) with the membrane-permeant Fe(II) and Fe(III) chelators BIP and PIH respectively. Moreover, the degree of dequenching caused by the sequential addition of these chelators (i.e. incubation with BIP followed by incubation with PIH) was used to determine relative differences in the redox state of the calcein-accessible fraction of the LIP. The reverse experiment (i.e. incubation with PIH followed by incubation with BIP) was not performed as PIH is capable of promoting the rapid oxidation of Fe(II) to Fe(III) , which would have confounded the interpretation of the results.
Prior to the initiation of an assay, cells were incubated for 10 min in pre-warmed (37°C) HBS containing 50 μM neocuproine. Calcein-fluorescence readings were performed in a FLUOstar Optima (BMG Labtech) microplate reader with excitation at 485 nm and emission detected at 520 nm. A stable baseline was established prior to the addition of ferric citrate by the addition of pre-warmed HBS containing 50 μM neocuproine. Ferric citrate (30 μM) was added where noted and the fluorescence followed for 10 min. Where noted, the dequenching of calcein fluorescence was monitored for 6 min following the addition of BIP (50 μM) and following the addition of PIH (50 μM). The majority of the fluorescence quenching was reversible with a combination of BIP and PIH (see Figure 5), confirming the specificity of the response for iron. Where indicated, the initial rates of fluorescence quenching are reported (typically obtained over the first 5 min following the addition of ferric citrate).
Ferrireduction assays were carried out as described previously . In brief, reduction of extracellular Fe(III) to Fe(II) by cultured astrocytes was assayed by exposing ascorbate-deficient or ascorbate-replete astrocytes to either 3 or 30 μM ferric citrate for 20 min at 37°C with orbital mixing at 320 rev./min. The ferene-S/Fe(II) complex was then determined colorimetrically in a microplate spectrophotometer [ferene-S/Fe(II); λmax = 593 nm; ϵ593 = 35.5 mM−1·cm−1] . Rates of cell-dependent ferrireduction were corrected for the rate of spontaneous ferrireduction that occurred in adjacent cell-free wells of the 24-well plate.
The protein content of astrocyte cultures was determined as described previously .
Statistical analysis and curve fitting
All statistical analyses were performed using GraphPad Prism® 5.0 (GraphPad Software). The dose–response data set for the ferristatin-mediated inhibition of 55Fe uptake was modelled using a sigmoidal dose–response curve with a Hill coefficient of 1 (‘Log[inhibitor] vs normalised response’ in GraphPad Prism). Differences between treatments were analysed using one-way ANOVA with Bonferroni's post hoc test of significance. Results are typically shown as means±SD. Significance levels are given in the Figure legends.
Astrocytes accumulate iron by an apparent Fe(III)-selective mechanism in the absence of ascorbate, but demonstrate enhanced iron accumulation from Fe(II) in the presence of ascorbate
We have investigated the accumulation of iron by astrocytes using two methods: (i) 55Fe transport assays from 55Fe-labelled ferric citrate; and (ii) intracellular calcein fluorescence quenching following exposure to ferric citrate. It should be noted that the term ‘iron accumulation’, as opposed to ‘iron uptake’, is employed here to allow for the possibility that some of the iron that is taken up by the astrocytes may be released during the course of the assays.
When exposed to 3 μM 55Fe-labelled ferric citrate for 1 h at pH 7.2 at 37°C, cultured rat astrocytes accumulated a mean value of 509±20 pmol iron/mg of protein (n=6). As shown in Figure 1(A), this accumulation of iron was diminished when the astrocytes were co-incubated with a 166-fold molar-excess (500 μM) of the membrane-impermeant Fe(II)-specific iron chelators, ferrozine (by 10%; P<0.05) and ferene-S (by 30%; P<0.001). Importantly, as similar concentrations of ferrozine typically inhibit the majority of iron uptake from low-molecular-mass ferric chelates by various cultured cell types [30–32], the present data are consistent with the notion  that cultured astrocytes are capable of importing iron from ferric–citrate complexes as Fe(III). Similarly, the addition of soluble Cp (caeruloplasmin; 60 μg/ml), which has potent ferroxidase activity  and can thus inhibit iron accumulation by cells in cases where there is a dependence of iron uptake on the reduction of Fe(III) to Fe(II) , significantly diminished iron accumulation by astrocytes, but only by 15% (Figure 1A). The similarity of the extent of inhibition of Cp with that of ferrozine and ferene-S suggests that Cp may be inhibiting the fraction of iron uptake by control astrocytes that is dependent on Fe(II) importation. Although there is a distinct possibility that the inhibitory effect of Cp on iron accumulation observed here (Figure 1A) could also be explained by Cp's capacity to enhance ferroportin-dependent iron-export , soluble Cp does not appear to stimulate iron release by astrocyte-like BT325 glioblastoma cells at concentrations of 25–300 μg/ml . Collectively, our results are supportive of the notion that when exposed to ferric citrate, cultured astrocytes are capable of Fe(III) uptake, but also indicate the presence of a small but significant level of Fe(II)-dependent uptake.
Astrocytes import Fe(III) and Fe(II) in the absence of ascorbate, but preferentially import Fe(II) in the presence of ascorbate
To assess the capacity for Fe(II)-dependent uptake, cultured astrocytes were presented with 3 μM 55Fe-labelled ferric citrate in the presence of a 500-fold molar excess (1.5 mM) of either L-ascorbate or the non-naturally occurring, but equally redox-active stereoisomer of L-ascorbate, D-isoascorbate. At a molar ratio of 500:1 (ascorbate/iron) the iron added as ferric citrate was predominantly in the Fe(II) form [>99.0% (results not shown)]. Both treatments resulted in a statistically significant doubling of the amount of cellular iron accumulated over 1 h. This accumulation was almost completely abolished by ferrozine (500 μM; Figure 1A), which is consistent with the likelihood that most of the iron had been reduced by ascorbate and was then bound to ferrozine and unable to be imported by either Fe(II)- or Fe(III)-dependent pathways. As expected, D-isoascorbate (‘D-AA’ in Figure 1A) also stimulated iron accumulation to a similar degree to L-ascorbate (Figure 1A), indicating that the stimulatory effect of extracellular ascorbate is not stereo-specific, and thus is likely to be mediated by the ascorbate-mediated reduction of Fe(III) to Fe(II).
To further strengthen our conclusions, the effects of extracellular ascorbate on iron accumulation were assessed using an iron-dependent calcein fluorescence-quenching assay (see the Experimental section). Calcein-loaded astrocytes were exposed to ferric citrate (30 μM) for 10 min in the presence or absence of L-ascorbate (1.5 mM), and the quenching of intracellular calcein fluorescence by imported iron was monitored. Although the fluorescence-quenching method employed can be directly related to cellular iron-uptake into the calcein-accessible component of the LIP , it is probably not directly related to cellular iron accumulation, the latter of which will include intracellular pools of iron that are not calcein-accessible. Nevertheless, the results obtained (Figure 1B) are consistent with the results of the 55Fe transport assays, and clearly indicate that the rate of intracellular fluorescence-quenching upon exposure to ferric citrate, which is indicative of cellular iron-import into the LIP [16,25,26], is markedly greater when Fe(III) is reduced to Fe(II) by a molar excess of ascorbate. Importantly, the presence of 1.5 mM ascorbate in the absence of exogenous iron (see the ‘baseline’ in Figure 1B and compare Figure 5A with Figure 5C) had no effect on intracellular fluorescence-quenching.
Collectively, the results from the present study strongly suggest that when in the presence of a molar excess of ascorbate (as is probably typical of the extracellular fluid of the brain) and low micromolar concentrations of iron citrate, cultured astrocytes accumulate iron at least twice as rapidly as ascorbate-deficient controls, with the additional iron being accumulated via a pathway that appears to be selective for Fe(II). Additionally, in the absence of ascorbate, cultured astrocytes appear to accumulate most of their iron by a mechanism that is consistent with the cellular uptake of Fe(III).
Ascorbate that is released from ascorbate-replete astrocytes stimulates iron reduction and accumulation
Astrocytes, similar to most mammalian cells (including those from non-scorbutic species), are unable to synthesize ascorbate de novo from glucose. Moreover, ascorbate is virtually absent from standard culture medium and demonstrates high lability under typical culture conditions [22,23]. Consequently, primary cultures of rodent astrocytes invariably demonstrate negligible levels of intracellular ascorbate [21,23,38].
Since the results above show that extracellular ascorbate enhances iron accumulation by cultured astrocytes, we sought to establish the effect of intracellular ascorbate on both astrocytic ferrireduction and iron accumulation. To achieve this aim, astrocytes were made replete with ascorbate by pre-exposure to DHA [12,21]. The exposure of ascorbate-deficient cultured astrocytes to 400 μM DHA for 30 min resulted in a mean intracellular ascorbate-loading of 38±5 nmol/mg of protein, consistent with values observed previously for ascorbate-loaded cultured astrocytes [21,23,38,39].
Although control astrocytes demonstrated low, but detectable, rates of ferrireduction, the loading of these cells with ascorbate stimulated their rates of ferrireduction by more than 6-fold (Figure 2A). Importantly, this stimulation was abolished by co-incubation with the ascorbate-oxidizing enzyme, AO (ascorbate oxidase) (Figure 2A), indicating that the stimulation of ferrireduction resulting from ascorbate-loading was entirely attributable to a direct reaction of extracellular Fe(III) with effluxed ascorbate . Additionally, ascorbate-replete astrocytes demonstrated a two-fold increase in both iron accumulation and intracellular calcein-quenching (Figures 2B and 2C respectively). It is worth noting here that, as with the results presented in Figure 1(B) (cf. Figure 5), ascorbate-loading itself did not affect intracellular calcein fluorescence, and a stable baseline was obtained before addition of ferric citrate (results not shown).
Effluxed ascorbate from ascorbate-replete astrocytes stimulates ferrireduction and iron accumulation
The observation that the co-incubation of ascorbate-replete astrocytes with ferene-S resulted in a near-complete inhibition of iron accumulation (Figure 2B) suggests, in line with the stimulation of iron accumulation obtained upon the addition of extracellular ascorbate (cf. Figure 1), that ascorbate loading stimulates iron accumulation by promoting the reduction of extracellular Fe(III) to Fe(II). Furthermore, the inhibitory effect of AO on both iron accumulation and intracellular calcein-fluorescence quenching (Figures 2B and 2C) demonstrates that most of the stimulatory effect of ascorbate loading can be attributed to the reduction of Fe(III) by effluxed ascorbate. Importantly, owing to the fact that AO completely abolished ascorbate-stimulated ferrireduction (Figure 2A), it is unlikely that the AO-insensitive fraction of iron accumulation is due to enzyme-catalysed tPMET (trans-plasma membrane electron-transfer) from intracellular ascorbate to extracellular Fe(III) . This AO-insensitive remainder of ascorbate-stimulated iron accumulation by cultured astrocytes remains to be explained. Additionally, the relative de-sensitization of iron accumulation to ferene-S in the presence of AO (Figure 2B) confirms that the inhibitory action of ferene-S is indeed due to the sequestration of Fe(II) formed by ascorbate-dependent ferrireduction.
Collectively, the results from the present study suggest that ascorbate loading stimulates Fe(II)-dependent iron accumulation by cultured astrocytes, and furthermore that this stimulation is largely mediated by the release of ascorbate into the extracellular space. Once released, this ascorbate presumably reacts directly with ferric citrate to form Fe(II) ions that can imported by transporters that are specific for Fe(II).
Iron accumulation by ascorbate-replete astrocytes can be inhibited by the DMT1 inhibitor ferristatin
As the results from the present study suggest that astrocytes show markedly enhanced iron accumulation in the presence of extracellular ascorbate (Figure 1), we next assessed the capacity of the DMT1 inhibitor ferristatin [16,18] to inhibit iron accumulation. This apparently competitive inhibitor blocks DMT1-dependent iron accumulation in DMT1-overexpressing HEK-293T cells [human embryonic kidney-293 cells expressing the large T-antigen of SV40 (simian virus 40)] with an IC50 of 14.7±1.5 μM at pH 6.8 when the cells are presented with 1 μM iron in the presence of 50 μM ascorbate .
We first established the dose–response relationship for ferristatin-mediated inhibition of iron accumulation from 55Fe-labelled ferric citrate by ascorbate-loaded astrocytes at an extracellular pH of 6.8. A pH of 6.8 was chosen for these experiments in order to allow a direct comparison between the IC50 data obtained in the present study and those obtained previously for DMT1-overexpressing HEK-293T cells . The results clearly indicate that ferristatin does inhibit iron accumulation in ascorbate-replete astrocytes (Figure 3). On the basis of these data, ferristatin was estimated to inhibit iron accumulation with an IC50 of 86±2 μM (95% confidence interval of the IC50 = 59–126 μM). However, if the iron accumulation data presented in Figure 3 are corrected for the component of iron accumulation that is insensitive to ferrozine [cf. Figure 1A; i.e. putative Fe(III)-selective uptake], with the resultant values thus representing putative Fe(II)-selective uptake, the estimated IC50 for ferristatin-mediated inhibition of iron accumulation from Fe(II) is 39±2 μM (95% confidence interval of the IC50 = 29–51 μM). Considering the 3-fold higher iron concentration employed in the present study when compared with that used by Buckett and Wessling-Resnick in their dose–response analyses , which would tend to increase the IC50 for a competitive inhibitor such as ferristatin, the presently observed dose–response relationship is consistent with the hypothesis that DMT1 provides a route of import for Fe(II) that is formed extracellularly by the reductive action of ascorbate on ferric citrate.
Dose–response of inhibition of iron accumulation by ferristatin at pH 6.8
We next assessed the capacity of ferristatin (30 μM) to inhibit iron accumulation by ascorbate-replete and ascorbate-deficient astrocytes at the more physiological pH of 7.2. Ferristatin inhibited both iron accumulation from 55Fe-labelled ferric citrate (3 μM; Figure 4A) and the initial rate of fluorescence quenching upon addition of ferric citrate (30 μM; Figure 4B) in both ascorbate-deficient and ascorbate-loaded astrocytes. The degree of inhibition of iron accumulation from 55Fe-labelled ferric citrate was markedly greater for ascorbate-replete than for ascorbate-deficient astrocytes (Figure 4A). This result is consistent with the view that ascorbate that has been released from astrocytes promotes ferrireduction and the accumulation of iron from Fe(II), and possibly does so without compromising that which is accumulated from Fe(III). This latter point provides some justification for the above subtraction of the ferrozine-insensitive component of astrocytic iron-accumulation from the enhanced rate (cf. Figure 3) observed in ascorbate-replete astrocytes.
Ferristatin inhibits iron accumulation by control and ascorbate-loaded astrocytes at pH 7.2
The effect of intracellular ascorbate on the redox state of labile iron in astrocytes
The present results suggest that cultured astrocytes demonstrate Fe(II)-dependent uptake in the presence of extracellular ascorbate, whether the latter is exogenously added to the extracellular compartment (Figure 1) or is released from cells (Figure 2). Moreover, we have presented pharmacological evidence suggesting that the accumulation of iron from Fe(II) by cultured astrocytes is largely DMT1-dependent (cf. Figures 3 and 4). In many cells, imported iron typically enters the cytosol as part of the LIP, which is considered to be a dynamic intracellular iron pool that is accessible to intracellular iron chelators such as calcein . As employed in the present study, the calcein-accessible component of intracellular iron can be monitored to detect iron influx into astrocytes. Moreover, as calcein chelates both Fe(II) and Fe(III), both of which quench fluorescence [27,40], the de-quenching of intracellular calcein-fluorescence by membrane-permeant iron chelators that are selective for either Fe(II) or Fe(III) can be used to assess the valency of chelated intracellular iron [25,37]. We have employed this approach to examine possible differences in the redox state of chelatable intracellular iron between ascorbate-deficient and ascorbate-replete astrocytes.
In the case of ascorbate-deficient astrocytes, the quenching of intracellular calcein fluorescence upon exposure of cells to 30 μM ferric citrate could not be reversed by the subsequent exposure of cells to the membrane-permeant Fe(II)-specific chelator BIP (50 μM; Figure 5A). However, quenched intracellular calcein-fluorescence was fully reversed by the application of the membrane-permeant Fe(III)-specific chelator PIH (50 μM; Figure 5A). On the other hand, in ascorbate-replete astrocytes, intracellular calcein-fluorescence that had been quenched upon exposure to ferric citrate was only fully reversed by the application of both BIP and PIH (Figure 5B). These results contrast with the situation observed in many other cultured cells grown under conditions of ascorbate deficiency in which the calcein-chelatable fraction of intracellular iron (resulting from iron uptake) is maintained in a form that is almost entirely ‘BIP-accessible’ [i.e. putatively in the Fe(II) state] . These data suggest that under the conditions employed in this study, the calcein-chelated fraction of intracellular iron resulting from NTBI uptake in cultured astrocytes is maintained: (i) as Fe(III) in ascorbate-deficient cells; and (ii) as both Fe(II) and Fe(III) in ascorbate-replete cells. Interestingly, the addition of 1.5 mM extracellular ascorbate to ascorbate-deficient astrocytes elicited calcein fluorescence quenching that was only reversible by PIH (Figure 5C). Thus although ascorbate-replete astrocytes maintain a LIP composed of both Fe(II) and Fe(III) (Figure 5B), ascorbate-deficient astrocytes, with (Figure 5C) or without (Figure 5A) 1.5 mM extracellular ascorbate, maintain a LIP composed almost entirely of Fe(III) (Figure 5C). It should be noted here that the rate at which cultured astrocytes become ascorbate loaded when exposed to extracellular ascorbate is typically much lower than when cells are exposed to DHA [14,21], and thus it would be expected that intracellular ascorbate levels in ascorbate-deficient astrocytes that have been treated with 1.5 mM ascorbate would be substantially lower than those treated with 400 μM DHA. Taking this into consideration, our data suggest that the appearance of Fe(II) within the LIP of cultured astrocytes is dependent on the accumulation of appreciable levels of ascorbate within the cytoplasm that are not likely to be achieved with short-term exposures to ascorbate.
Ascorbate-replete astrocytes demonstrate an apparent shift in the valency of the intracellular LIP of imported iron
It is worth noting that in Figure 5(C), the addition of BIP to astrocytes incubated in the presence of 1.5 mM ascorbate in fact stimulates further calcein fluorescence-quenching (Figure 5C). This may suggest that the chelation of intracellular Fe(II) by BIP stimulates further entry of iron into the LIP, which is subsequently bound by calcein.
Most mammalian cells examined appear to demonstrate a strong preference for importing NTBI in its ferrous form . This preference is supported by the observation that membrane-impermeant Fe(II) chelators such as ferrozine [31,32] substantially inhibit (by 50–95%) the accumulation of iron from NTBI when the iron is supplied to cells in a ferric form. It is also supported by findings from numerous groups that DMT1, which transports Fe(II), but not Fe(III), is currently the only confirmed transporter of NTBI in mammalian tissues (for a review see ). In astrocytes, however, it remains unclear in which valency the importation of NTBI occurs [1,9]. The present study investigated the astrocytic accumulation of iron from Fe(II) and Fe(III) using a range of new approaches. For the first time in an investigation of iron uptake by astrocytes, this study has: (i) examined the effect of loading cells with ascorbate on iron accumulation; (ii) examined the extent to which iron accumulation can be prevented by a pharmacological inhibitor of DMT1; and (iii) used membrane-permeant chelators that are either selective for Fe(II) or Fe(III) to examine the valency of the intracellular LIP.
The present study has also used ferric citrate, rather than the frequently used FAC. Citrate is present within the extracellular fluid of the brain , and since citrate is an effective iron chelator, ferric citrate is a likely source of NTBI in the healthy brain . The present study suggests for the first time that ascorbate that is released from ascorbate-replete astrocytes markedly stimulates their rates of iron accumulation from NTBI, and significantly advances a recent report that cultured astrocytes are capable of accumulating iron from Fe(II) .
The results obtained in the present study, combined with previous observations , provide further evidence for two separate routes of NTBI uptake in astrocytes. The first route is markedly enhanced by the reduction of Fe(III) to Fe(II) by ascorbate that has been released from the cell interior, followed by the importation of Fe(II) by DMT1; whereas the other route is apparently independent of both ascorbate and DMT1, and may involve the import of Fe(III). Our results are consistent with the hypothesis that in ascorbate-replete astrocytes both the Fe(II) and Fe(III) importation pathways make a similar contribution to the overall amount of iron accumulated, whereas in ascorbate-deficient astrocytes most of the iron is taken up as Fe(III). The implications of these findings are discussed below.
Tulpule and colleagues recently suggested that cultured astrocytes are capable of importing both Fe(II) and Fe(III), and noted that the apparent rates of ferrireduction by astrocytes cultures are markedly less than the rates of iron accumulation from 100 μM FAC (3.2±0.4 nmol/h per mg of protein compared with 24.7±8.9 nmol/h per mg of protein respectively) . This difference is consistent with the notion that ascorbate-deficient astrocytes import most of their NTBI as Fe(III), although it remains possible that the detected rates of Fe(II) formation in the previous study underestimate actual Fe(II) generation at the cell surface. This might be expected to occur if there was a high-affinity metabolic-coupling between the ferrireductase and the Fe(II) transporter. A similar criticism can also be lodged against the use of ferrozine in the present study to establish putative Fe(III)-selective iron importation (Figure 1A). However, the observation that excess ferrozine almost completely inhibits the importation of iron from extracellular Fe(III) by other cell types [31,32], suggests that sufficiently high concentrations of membrane-impermeant chelators, such as ferrozine, can chelate almost all Fe(II) that is generated extracellularly. The present study has examined the apparent rates of iron accumulation from 55Fe-labelled ferric citrate in the presence of a large molar-excess of the membrane-impermeant Fe(II) chelators, ferrozine and ferene-S. Iron accumulation under these conditions was slowed by 10–30%, which is comparable to the 13% reported previously , and suggests that cultured astrocytes are indeed capable of importing iron as both Fe(III) and Fe(II).
Consistent with the recent suggestion that astrocytes are capable of Fe(II)-selective uptake , we similarly observed that a large excess of extracellular ascorbate increases the rate of iron accumulation. Moreover, we have for the first time observed that the loading of astrocytes with ascorbate stimulates their accumulation of iron to a similar degree, and furthermore that this stimulation depends on the reduction of extracellular Fe(III) by ascorbate that is released by the cells. These results suggest that when replete with ascorbate, astrocytes are able to accumulate at least half of their iron by an importation route that is selective for Fe(II). These results are consistent with our previous findings that cellular ascorbate-efflux from ascorbate-replete K562 cells enhances their iron accumulation from NTBI , and further suggests that the involvement of cellular ascorbate release in NTBI uptake may be ubiquitous.
It has recently been demonstrated that cultured astrocytes express mRNA for two putative ascorbate-dependent cytochromes b561, Dcytb and SDR2 . Although it is possible that these enzymes could be responsible for mediating the stimulatory effect of ascorbate on ferrireduction and iron accumulation, our results do not support this hypothesis for the present study, which predicts that there should be a component of the enhanced rate of ferrireduction that is insensitive to the membrane-impermeant enzyme, AO. On the contrary, we observed that the enzyme completely inhibited the stimulated component of cellular ferrireduction that was obtained after astrocytes were loaded with ascorbate (Figure 2A). These data suggest that although ascorbate-stimulated iron accumulation by astrocytes involves extracellular ferrireduction by effluxed ascorbate, it does not involve electron donation directly to extracellular Fe(III) by ascorbate-dependent transplasma membrane ferrireductases. However, it should be noted that the results from the present study remain consistent with the recent hypothesis  that cytochromes b561 may stimulate extracellular ferrireduction by transferring electrons from intracellular ascorbate to an extracellular ascorbate intermediate (e.g. the ascorbyl radical), which is analogous to the well-established mechanism of cytochrome b561-catalysed electron transfer across chromaffin granule membranes.
The present study also provides the first pharmacological evidence that DMT1 may be responsible for a large fraction of Fe(II) uptake by cultured astrocytes. These results were obtained using the DMT1 inhibitor ferristatin . This inhibitor is a poly-sulfonated dye containing two copper centres . As only one of these copper centres appears to be labile and thus could potentially interfere with iron uptake measurements , all iron uptake assays were carried out in the presence of the membrane-permeant copper chelator neocuproine (see the Experimental section) instead of the previously used triethylenetetramine . Ferristatin appears to inhibit DMT1-dependent iron uptake in DMT1-overexpressing HEK-293T cells with an IC50 of approx. 15 μM , whereas we have observed that ferristatin inhibits iron accumulation by cultured rat astrocytes with an apparent IC50 of 86±2 μM at pH 6.8. However, this latter IC50 value is confounded by the fact that astrocytes appear to take up appreciable amounts of Fe(III). If we assume that the uptake of Fe(III) uptake is unaffected by the presence of ascorbate, the adjusted IC50 for ‘Fe(II)-selective’ iron accumulation is somewhat lower (39±2 μM). Importantly, as DMT1 only transports iron as Fe(II) [3,10], the observation in the present study of an apparent DMT1-dependence of iron accumulation in the presence of ascorbate supports the existence of an Fe(II) import pathway in cultured astrocytes.
Our results from the present study suggest that the uptake of Fe(II) via DMT1 can occur in astrocytes at an extracellular pH of 7.2, which is thought to correspond to the pH of the extracellular fluid in normal brain tissue. Although DMT1 has proton-symport activity at acidic pH values, with the maximal transport rate of DMT1 occuring at an extracellular pH of ~5.5, significant transport rates are still observed at pH values of 7.0–7.4 [10,42]. Indeed, it has been demonstrated that at an extracellular pH of 7.4, DMT1 demonstrates an Fe(II) conductance activity that is not coupled to proton symport .
The mechanism for astrocytic Fe(III) uptake observed in the present study remains unknown. It was noted previously  that two possibilities for Fe(III)-selective transport include the elusive β3-integrin/mobilferrin  and trivalent-cation-selective  iron uptake pathways, both of which have yet to be examined in astrocytes. Although an interaction of Tf with the TfR1 (Tf receptor 1) could potentially explain the observation of Fe(III)-selective uptake in the present study, this possibility is highly unlikely. First, the extensive washing of the cells prior to iron-accumulation experiments with serum-free HBS (a total of 6 individual washes, before and after pre-incubations) would have removed virtually all bovine Tf remaining after removal of the growth medium. Secondly, all cells underwent a 30 min pre-incubation at 37°C before iron-accumulation experiments, and it is known that >90% of internalized Tf is typically released within 30 min at this temperature . Thus, by the time that cells were used in iron accumulation experiments, it is reasonable to assume that virtually all exogenous Tf had been removed. Finally, although astrocytes have been shown to be able to synthesize and secrete Tf , only very low levels (~50 ng/ml) are secreted over a period of several days . Thus, as all iron accumulation experiments in the present study were conducted over a period of 60 min, it is highly unlikely that any endogenous Tf that may have been secreted from the astrocytes during this time would have significantly contributed to the Fe(III)-selective uptake observed in our experiments.
The extent to which the presently described NTBI uptake pathways contribute to iron uptake by astrocytes in vivo remains to be determined. In many cells, the contribution of NTBI uptake is typically small compared with the contributions made by TBI uptake. However, in the case of the brain, there are indications that the relative contributions made by these iron uptake pathways may be quite different [2,5]. Although astrocytes in culture express TfR1 , TfR1 may not be expressed on astrocytes in vivo, at least under normal conditions . This suggests that astrocytes in vivo may not normally utilise Tf-bound iron as a source of the metal . Indeed it has recently been suggested that although neurons clearly express TfR1  and are consequently thought to employ TBI uptake in vivo , astrocytes and other glial cells (e.g. microglia and oligodendrocytes) may rely more heavily on NTBI uptake . As there is mounting evidence that a substantial fraction of the NTBI uptake by cultured astrocytes is imported as Fe(III), there is now strong justification for investigating the mechanism(s) responsible for this mode of uptake.
The present study also found that the process of loading astrocytes with ascorbate shifts the redox state of the chelatable fraction of subsequently accumulated iron from almost entirely Fe(III) to a combination of Fe(II) and Fe(III). One interpretation of these results is that intracellular iron is maintained in the short term in the same valence state in which it was imported. This interpretation implies the presence of intracellular chaperones or chelators that are capable of binding Fe(II) and Fe(III) to prevent them from shifting oxidation state. Against this interpretation, however, is our observation that when ascorbate-deficient astrocytes are exposed to extracellular ascorbate (1.5 mM), most of the calcein-chelated fraction of intracellular iron is present as Fe(III) (Figure 5C). Therefore our results suggest the alternative hypothesis that the differences in the valency of calcein-chelated intracellular-iron, before and after ascorbate-loading, are due to the reductive influence of intracellular ascorbate. The potential of intracellular ascorbate to influence the redox state of the intracellular LIP in cultured astrocytes, and potentially in other cell types , warrants further investigation; particularly given that cells grown under standard culture conditions are chronically ascorbate-deficient .
In conclusion, we have provided further evidence that cultured rat astrocytes are capable of accumulating iron from Fe(II), and have provided support for the notion that cultured astrocytes are capable of importing Fe(III). The uptake of Fe(III) appears to predominate in the absence of ascorbate, whereas the uptake of Fe(II) appears to be responsible for at least half of the iron accumulated when astrocytes are replete with ascorbate. This Fe(II) uptake appears to depend largely on ferrireduction by extracellular ascorbate and on the activity of DMT1. The mechanism of Fe(III) uptake remains uncertain. Moreover, intracellular ascorbate appears to shift the redox state of intracellular chelatable-iron from Fe(III) to a mixture of Fe(II) and Fe(III). As ascorbate is present at concentrations of 200–400 μM in the extracellular fluid of the brain, and at millimolar concentrations within cells , it is likely to be intimately involved in the accumulation of iron by astrocytes in the mammalian brain.
calcein acetoxymethyl ester
divalent metal transporter 1
ferric ammonium citrate
cells, human embryonic kidney-293 cells expressing the large T-antigen of SV40 (simian virus 40)
labile iron pool
pyridoxal isonicotinoyl hydrazone
transferrin receptor 1
Darius Lane designed, performed and analysed all experiments, and drafted the manuscript. Stephen Robinson and Glenda Bishop supervised Hania Czerwinska. Stephen Robinson was involved in the writing of the manuscript. Hania Cerwinska was involved in the preparation and maintenance of primary astrocytes. Alfons Lawen supervised Darius Lane, was involved in the analysis of all experiments, and supervised the writing of the manuscript.
We wish to thank the NIH National Cancer Institute for the supply of ferristatin, Professor D. Richardson for the gift of PIH and Professor M. Wessling-Resnick for advice on the use of ferristatin.
D. J. R. L. and A. L. acknowledge the financial support of the Department of Biochemistry and Molecular Biology, Monash University. D. J. R. L. was a recipient of a Faculty of Medicine, Nursing and Health Sciences (Monash University) bridging fellowship. S. R. R. and G. M. B. received financial support from a National Health and Medical Research Council Project Grant [grant number 334219] and from the School of Psychology and Psychiatry, Monash University.