Iron induces two distinct Ca2+ signalling cascades in astrocytes

Iron is the fundamental element for numerous physiological functions. Plasmalemmal divalent metal ion transporter 1 (DMT1) is responsible for cellular uptake of ferrous (Fe2+), whereas transferrin receptors (TFR) carry transferrin (TF)-bound ferric (Fe3+). In this study we performed detailed analysis of the action of Fe ions on cytoplasmic free calcium ion concentration ([Ca2+]i) in astrocytes. Administration of Fe2+ or Fe3+ in μM concentrations evoked [Ca2+]i in astrocytes in vitro and in vivo. Iron ions trigger increase in [Ca2+]i through two distinct molecular cascades. Uptake of Fe2+ by DMT1 inhibits astroglial Na+-K+-ATPase, which leads to elevation in cytoplasmic Na+ concentration, thus reversing Na+/Ca2+ exchanger and thereby generating Ca2+ influx. Uptake of Fe3+ by TF-TFR stimulates phospholipase C to produce inositol 1,4,5-trisphosphate (InsP3), thus triggering InsP3 receptor-mediated Ca2+ release from endoplasmic reticulum. In summary, these findings reveal the mechanisms of iron-induced astrocytic signalling operational in conditions of iron overload. Wenzheng Guan and Maosheng Xia et al. use a combination of pharmacological, genetic, and calcium imaging approaches to describe two mechanisms of iron-induced astrocytic calcium signaling. Altogether, these results suggest that astrocytes may act as a barrier to excessive iron in the brain, thereby improving our understanding of the role of astrocytes in normal neurodevelopment.

I ron contributes to numerous cellular and biochemical processes and acts as a co-factor in various molecular cascades in the nervous tissue including the synthesis and metabolism of several brain-specific enzymes and neurotransmitters 1,2 . In biological systems iron is present in either reduced ferrous (Fe 2+ ) or oxidized ferric (Fe 3+ ) state. The brain has the second (after liver) highest quantity of iron in the human body with total non-haem iron in the brain reaching about 60 mg 3 . The non-haem iron concentration in the serum ranges between 9 and 30 μM, whereas the iron concentration in cerebrospinal fluid (CSF) is much smaller being around 0.3-0.75 μM 4,5 . Transport of iron across the blood-brain barrier (BBB) is mediated either by transferrin receptor (TFR)-mediated endocytosis of Fe 3+ -bound to transferrin (holo-TF), or, for non-TF-bound iron, by vesicular and non-vesicular pathways 6 . Membrane transport of Fe 2+ is also mediated by divalent metal ion transporter 1 (DMT1/SLC11A2) which underlies Fe 2+ uptake through the plasma membrane or from endosomes 6 . Under physiological conditions, the intracellular cytosolic ionised iron levels fluctuate around 0.5-1.5 μM 7 .
In the brain, up to three-fourths of total iron is accumulated within neuroglial cells 8 . Astrocytes in particular are fundamental elements of ionostatic control over CNS environment 9 . Ionised Fe 2+ enters astrocytes through DMT1/SLC11A2 transporters which are particularly concentrated in the endfeet of cerebral and hippocampal astrocytes 10 . In physiological condition, DMT1 is widely distributed in the nervous system, being expressed in neurones [11][12][13] , astrocytes [13][14][15] and oligodendrocytes 14,16 in vitro and in the brain tissue. In astrocytes, DMT1 mediates nontransferrin-bound iron (NTBI) transport, thus contributing to the brain iron homoeostasis in development and adulthood [17][18][19] . Conditional deletion of DMT1 form oligodendrocyte precursors substantially inhibited both myelination in development and remyelination in pathology 16 . Evidence for the expression of TFR in astroglial cells remains controversial 6,20,21 , while iron overload may influence the expression or distribution of TFR in astrocytic compartments 21,22 . Cellular uptake of Fe 3+ requires internalization of TF-TFR complex 23 . An adaptor protein Disabled-2 (Dab2) plays an essential role in cell signalling, migration and development 24 . In mice the Dab2 has two isoforms of 96 and 67 kDa (p96 and p67 24 ). In human K562 cells, Dab2 regulates internalization of TFR and uptake of TF 25 . Dab2 is also widely distributed in immune cells and in neuroglia 24 , although the functional link between Dab-2 and TFR in astrocytes has not been demonstrated.
Astrocytes possess a special form of intracellular ionic excitability, mediated by temporal and spatial fluctuations in the intracellular ion concentration 26,27 . Astroglial Ca 2+ signalling is mediated by Ca 2+ release from the endoplasmic reticulum (ER) following activation of inositol-1,4,5-trisphosphate receptor (InsP 3 R), or intracellular Ca 2+ -gated Ca 2+ channels known as ryanodine receptors (RyR). Astroglial Ca 2+ signals may also be generated by plasmalemmal Ca 2+ entry through Ca 2+ -permeable channels or by sodium-calcium exchanger (NCX) operating in the reverse mode 26,28 . Astroglial Na + signalling is shaped by plasmalemmal Na + entry through cationic channels and numerous Na + -dependent transporters, of which the major role belongs to Na + -dependent glutamate transporters [29][30][31] ; Na + extrusion is mediated by the sodium-potassium pump (NKA). Both Na + and Ca 2+ signalling systems are closely coordinated, with NKA and NCX accomplishing this coordination at the molecular level 32 . Astrocytes specifically express α2-subunit containing NKA which is fundamental for astroglial K + buffering 33 . Astrocytes express all three isoforms of NCX -NCX1/SLC8A1, NCX2/SLC8A2 and NCX3/SLC8A3, with some evidence indicating higher expression of NCX1 34 . The NKA, the NCX and glutamate transporters are known to be preferentially concentrated in the perisynaptic astroglial membranes indicating intimate relationship between these ion-transporting molecules 35,36 . The NCX is also known to localise at caveolae rich in caveolin-3 (Cav-3), the latter isoform being predominantly expressed in astrocytes 37 .
In the present paper we performed an in depth analysis of the action of ferrous and ferric (Fe 2+ and Fe 3+ ) on astrocytic Ca 2+ and Na + dynamics. We found that Fe 2+ (through DMT1) and Fe 3+ -TF (through TFR) evoke [Ca 2+ ] i transients in astrocytes in culture and in vivo. Effects of Fe 2+ on [Ca 2+ ] i are mediated mainly by the reversed NCX, whereas Fe 3+ triggers Ca 2+ release from the endoplasmic reticulum by stimulations of InsP 3 R. In conclusion, we discovered that iron ions trigger astrocytic Ca 2+ signalling by acting through two distinct molecular cascades.

Results
Fe 2+ /Fe 3+ trigger [Ca 2+ ] i increase in cortical astrocytes in vitro and in vivo. Exposure of cultured astrocytes to Fe 2+ leads to a gradual increase in cytoplasmic Fe 2+ concentration as revealed by quenching of Fura-2 (Supplementary Figure 1). Washout of Fe 2+ leads to a full recovery of Fura-2 fluorescence; these data indicate the existence of plasmalemmal Fe 2+ transport system as well as absence of non-specific damaging effect of ionised iron on the cellular membrane. We analysed effects of Fe 2+ and Fe 3+ Ca 2+ dynamics in astrocytes in primary cultures and in vivo in GFAP-eGFP transgenic mice (Fig. 1 (Fig. 1a).
DMT1 and TFR mediate Fe 2+ and Fe 3+ uptake. As mentioned above, Fe 2+ uptake could be mediated by plasmalemmal transporter DMT1, whereas Fe 3+ is mainly accumulated in TF-bound form by TFRs (Fig. 2a). Immunostaining of cortical tissue preparations and primary cultured astrocytes demonstrated colocalisation of DMT1 and TFR with astroglial GFAP-positive profiles (Fig. 2b). In the cortical tissue both DMT1 and TFR are preferentially localised at perivascular endfeet. Meanwhile, expression of specific DMT1 and TFR mRNA was also detected in the freshly isolated and sorted astrocytes and neurones, as well as and in the cerebral tissue (Fig. 2c).
To reveal the contribution of DMT1 and TFR to Fe 2+ / Fe 3+ -induced [Ca 2+ ] i dynamics, we inhibited expression of DMT1 or TFR using siRNA duplex chains. The representative western blots demonstrating the efficacy of knockdown are shown in Fig. 2d. When compared to the control group, the DMT1 siRNA reduced expression of DMT1 to 9.52% ± 2.58% (n = 6, p < 0.0001), whereas treatment with TFR siRNA decreased TFR levels to 7.84% ± 2.10% (n = 6, p < 0.0001) from the control values. Administration of Fe 2+ to DMT1-deficient astrocytes failed to induce any changes in [Ca 2+ ] i . At the same time Fe 2+ induced robust [Ca 2+ ] i elevation in astrocytes treated with siRNA(−) (Fig. 2e). Similarly, Fe 3+ did not produce [Ca 2+ ] i , transients in astrocytes treated with TFR siRNA duplex chains, whereas in cells exposed to negative control siRNA Fe 3+ evoked [Ca 2+ ] i elevations (Fig. 2e). At the same time Fe 3+ -induced [Ca 2+ ] i transients were fully preserved in DMT1-deficient astrocytes (Supplementary Figure 2), hence questioning the role of DMT1 in release of Fe 3+ from endosomes.
In unstimulated astrocytes in culture the DMT1 fluorescence was the highest around the nucleus, suggesting its preferred intracellular localisation. After treatment of cultures with Fe 2+ for 5 min we observed redistribution of DMT1 from the nuclear region to the plasma membrane (Fig. 2f). The levels of DMT1 were measured in the extracted proteins of the nucleus and cytoplasm (Fig. 2g). After treatment with Fe 2+ for 5 min, the level of DMT1 in the nuclei decreased to 36.41% ± 4.33% of control group (n = 10, p < 0.0001), whereas the level of cytoplasmic DMT1 increased to 117.37% ± 5.79% of control group (n = 10, p < 0.0001) (Fig. 2g).
Sources of iron-induced [Ca 2+ ] i mobilisation. The main sources for [Ca 2+ ] i increase in astrocytes are (i) Ca 2+ release from the ER following the opening of InsP 3 Rs or RyRs, or (ii) plasmalemmal Ca 2+ influx through either Ca 2+ permeable channels (such as L-type Ca 2+ channels or TRP channels) or NCX operating in the reverse mode (Fig. 3a) (Fig. 3c). In contrast, inhibition of NCX with Fig. 1 Iron ions, Fe 2+ and Fe 3+ , evoke astrocytic intracellular Ca 2+ signals in vitro and in vivo. a Representative images (top panel) and intracellular Ca 2+ ([Ca 2+ ] i ) recordings from primary cultured astrocytes in response to different concentrations of Fe 2+ or Fe 3+ . Every data point represents mean ± SD, n = 10, p < 0.05, statistically significant difference from the value of baseline in the same group. The experiment was repeated in 10 different cultures. The representative image on the left shows the primary cultured astrocytes co-stained with GLT1 (green), GFAP (red), and DAPI (blue). Representative Fluo-4 images (green) show astrocytes treated by 10 μM Fe 2+ or Fe 3+ at 0 s and 105 s of the recordings, the astrocytic identity was confirmed by costaining with SR101. Scale bar, 20 μm. b Images (top panel) and [Ca 2+ ] i recordings from cortical astrocytes in GFAP-eGFP transgenic mice using transcranial confocal microscopy. Every data point represents mean ± SD, n = 10 (different mice), p < 0.05, statistically significant difference from the value of baseline in the same group. Representative Rhod-2 images (red) show astrocytes treated with 100 μM Fe 2+ or Fe 3+ at 0 s (baseline) and 45 s (peak of the response); GFAP-eGFP images (green) are shown on the right. Scale bar, 10 μm.  (Fig. 3f).
DMT1 transports Fe 2+ , which inhibits NKA, increases [Na + ] i and reverses NCX. Experiments described above have demonstrated that Fe 2+ , after being transported into the cell by DMT1, leads to a reversal of the NCX, which results in Ca 2+ influx. The NCX reversal in astrocytes is triggered by an increase in [Na + ] i . Such an increase may originate either from the activation of plasmalemmal Na + entry or from inhibition of the NKA, which maintains basal [Na + ] i 28,30 . The activity of NKA was suppressed by exposure to 10 μM Fe 2+ to 82.40 ± 5.74% (n = 10, p < 0.0001) of the control. Exposure to 100 nM of the specific NKA inhibitor, ouabain, reduced NKA activity to 72.30 ± 5.91% (n = 10, p < 0.0001) of the control (Fig. 4a). When 10 μM Fe 2+ and 100 nM ouabain were added together, the NKA activity was reduced to 71.80 ± 7.81% (n = 10, p < 0.0001) (Fig. 4a).
Inhibition of NKA in astrocytes results in a substantial elevation in [Na + ] i . When monitoring [Na + ] i in cultured astrocytes with Na + -sensitive probe SBFI we found that both Fe 2+ (10 μM) and ouabain (100 nM) triggered rapid and substantial elevation of [Na + ] i (Fig. 4b). These changes in [Na + ] i were paralleled by [Ca 2+ ] i dynamics. Exposure of astrocytes to Fe 2+ , or ouabaine or mixture of Fe 2+ and ouabain caused [Ca 2+ ] i elevation (Fig. 4c). When Fe 2+ was applied in the presence of ouabain it failed to change [Ca 2+ ] i (Fig. 4d); at the same time application of Fe 3+ in the presence of ouabain still triggered additional [Ca 2+ ] i elevation (Fig. 4d).
Fe 2+ -induced Ca 2+ mobilisation is associated with caveolae. Treatment of cultured astrocytes with interfering Cav3 siRNA duplex chains decreased the level of Cav3 to 7.41 ± 4.32% (n = 6, p < 0.0001) of the control (Fig. 4e). An in vitro knock-down of Cav3 significantly reduced amplitudes of Fe 2+ -induced [Ca 2+ ] i Fig. 2 Astrocytic expression of DMT1 and TFR. a Astrocytes accumulate Fe 2+ through plasmalemmal divalent metal transporter 1 (DMT1/SLC11A2) whereas Fe 3+ is taken up by internalisation of Fe 3+ -TF-TFR complex. b Images of astrocytes in the somato-sensory cortex preparations or in primary culture double-immunolabeled with antibodies against DMT1 or TFR and against GFAP. Scale bar, 20 μm. c The mRNA expression of DMT1 and TFR measured by qPCR in astrocytes sorted from GFAP-GFP mice, in neurones sorted from Thy1-YFP mice, and in whole cerebral tissue of wild type mice. d Representative western blot bands for DMT1 and TFR in cultured astrocytes treated with sham (Control), siRNA negative control (−) or positive duplex chains (+). The protein levels are shown as the ratio of DMT1 (55 kDa) and β-actin (42 kDa), and TFR (92 kDa) and β-actin. Data represent mean ± SD, n = 6. *Indicates statistically significant (p < 0.05) difference from the value of baseline in the same group. e [Ca 2+ ] i transients evoked by Fe 2+ or Fe 3+ after RNA interference and down-regulation of protein synthesis. After treatment with DMT1 or TFR siRNA negative control (−) or positive duplex chains (+) for 3 days, [Ca 2+ ] i dynamics i n response to Fe 2+ /Fe 3+ was monitored. Every data point represents mean ± SD, n = 10. f Images of astrocytes in primary culture treated with sham (Control) or Fe 2+ for 5 min and double-immunolabelled with antibodies against DMT1 and against GFAP. Scale bar, 25 μm. g Redistribution of DMT1 induced by Fe 2+ in the extracted proteins of nucleus and cytoplasm. Every data point represents mean ± SD, n = 10. *Indicates statistically significant (p < 0.05) difference from control group.
When Dab2-deficient astrocytes were challenged with Fe 3+ , no [Ca 2+ ] i increase was recorded (Fig. 6d). Similarly, after inhibition of the PLC with U-73122, application of Fe 3+ did not change [Ca 2+ ] i (Fig. 6d). Hence, we may surmise that uptake of Fe 3+ through TFR requires Dab2 protein; after entering the cytosol Fe 3+ activates the PLC, which produces InsP 3 that triggers InsP 3induced Ca 2+ release from the ER (Fig. 6e).

Discussion
In this paper, we describe previously unknown effects of iron ions on cellular [Ca 2+ ] i in astrocytes. Administration of either Fe 2+ or Fe 3+ triggered a concentration-dependent increase in [Ca 2+ ] i with EC 50 of about 0.4-0.6 μM. We performed an in depth analysis of the mechanisms underlying iron transport and iron induced Ca 2+ signalling. We also demonstrated that, contrary to the previous beliefs, astrocytes express functional TFR in vitro and in vivo thus allowing accumulation of Fe 3+ .
In the brain, majority of cells express a full complement of proteins responsible for iron homoeostasis, including TFR and DMT1 for iron uptake, heavy and light chains ferritin for iron b Protein levels are shown as the ratio of DMT1 (55 kDa) and β-actin (42 kDa), NCX1 (108 kDa) and β-actin, NCX2 (102 kDa) and β-actin, NCX3 (100 kDa) and β-actin, NKA-α1/α2 (100 kDa) and β-actin (42 kDa). Data represent mean ± SD, n = 6. *Indicates statistically significant (p < 0.05) difference from any other group; **indicates statistically significant (p < 0.05) difference from control plus Fe 2+ group or Cav3 siRNA(+) plus Ctrl group. sequestration, cytosolic iron exporter ferroportin 1 (FPN1), and iron regulatory protein 1 and 2 (IRP1and IRP2) for regulating intracellular iron homoeostasis 39,40 . Glial cells, and astrocytes in particular, store up to 75% of ionised iron in the CNS 41 , and protect the brain against iron overloads 42 . Transmembrane transport of iron in astrocytes has been identified, but was not studied in details. There is a general agreement on the primary role of plasmalemmal divalent metal transporter 1, DMT1/ SLC11A2, which selectively transports Fe 2+ ; the DMT1 was detected in astrocytes in culture and there is evidence indicating its presence in astroglial endfeet in situ [43][44][45] . In addition, Fe 2+ was suggested to enter reactive astrocytes by diffusion through transient receptor potential canonical (TRPC) channels 42 . Expression of TF-Fe 3+ -transporting TFR has been noted in astrocytes in culture 21,46 ; it is, nonetheless, generally believed that astrocytes in vivo are not in a possession of TFR and hence cannot accumulate Fe 3+ 47-49 . This conclusion, however, has been made on the basis of rather limited investigations 14,46 ; while expression of TFR-specific mRNA was detected in astrocyte transriptome 50 . In our study we confirmed expression of DMT1, at mRNA and protein levels as well as by immunostaining, in acutely isolated astrocytes, in astroglial primary culture and in situ in cortical tissue; the DMT1 was particularly enriched in the endfeet (Fig. 2b-d). We also detected astroglial expression of TFR at mRNA level in the transcriptome of acutely isolated and FACS-sorted astrocytes (Fig. 2c). We further confirmed expression of TFR in astrocytes at a protein level and in immunohistochemical analysis of astrocytes in culture and in cortical preparations ( Fig. 2b-d). In the cortical tissue TFR labelling was concentrated in perivascular astrocytic endfeet (Fig. 2b-d).
Not much is known about the links between ionised iron and Ca 2+ signalling in the cellular elements of the CNS. In the literature, we found only a single example of Fe 3+ (Fig. 2). The Fe 2+ -induced [Ca 2+ ] i Protein values are shown as the ratio of 96 kDa isoform and β-actin (42 kDa), and the ratio of 67 kDa isoform and β-actin. Data represent mean ± SD, n = 6. *Indicates statistically significant (p < 0.05) difference from any other group. c Fe 3+ -dependent changes in the InsP 3 after treatment with Dab2 siRNA duplex chains. After treatment with Dab2 siRNA duplex chains (+) for 3 days, the primary cultured astrocytes were treated with serum-free medium (Control), TF (as negative control) or 10 μM Fe 3+ -TF (Fe 3+ ) for 5 minutes, the level of InsP 3 was measured by ELISA and shown as mean ± SD, n = 10.  (Fig. 3). These data indicate that Fe 2+ , after being accumulated in the astrocyte, switches the NCX into the reverse mode of operation thus generating Ca 2+ influx into the cell in exchange for Na + . This scenario requires increase in astroglial [Na + ] i , which readily reverses the NCX 28,52 . An increase in [Na + ] i is likely to result from the inhibition of NKA, which represents the major Na + efflux mechanism in astrocytes 30 . Activity of NKA was effectively suppressed by Fe 2+ , and probing astrocytes with Na + -sensitive indicator SBFI revealed Fe 2+ -induced [Na + ] i elevation (Fig. 4). In the presence of NKA inhibitor ouabain Fe 2+ -induced [Ca 2+ ] i responses were eliminated thus corroborating the central role of NKA and NCX in Fe 2+ -induced Ca 2+ signalling (Figs. 4, 7). Of note, exposure to Fe 2+ initiated rapid redistribution of DMT1 from nucleus into cytoplasm (Fig. 2g) and arguably to the plasmalemma thus increasing astrocyte capacity for iron uptake.
The mechanism of Fe 3+ -induced Ca 2+ signalling is associated with intracellular Ca 2+ release. The Fe 3+ -induced [Ca 2+ ] i responses are preserved in Ca 2+ -free extracellular solution while being blocked by XeC (inhibitor of InsP 3 receptor) and by U-73122 (inhibitor of PLC) thus revealing the central role for InsP 3medaited ER Ca 2+ release. Initiation of this signalling cascade requires transmembrane transport of Fe 3+ : in vitro knockdown of TFR eliminates Fe 3+ -evoked [Ca 2+ ] i dynamics. The internalisation of TF-Fe 3+ -TFR complex also requires functional Dab2 protein. This protein is a multi-modular scaffold protein with signalling roles in ion homoeostasis, inflammation and receptors internalization 53 . Ablation of Dab2 in astrocytes with specific siRNA interrupts signalling chain and blocks Fe 3+ -dependent Ca 2+ signalling (Figs. 6 and 7). Import of TF-Fe 3+ -TFR complex involves endocytosis; acidic (pH~5.6) environment of the endosome achieved through the action of ATP-dependent H + pumps 54 , reduces the affinity of TF to Fe 3+ and the latter is released 55 . Subsequently, an endosomal reductase reduces Fe 3+ to Fe 2+ , and Fe 2+ can be transported into the cytosol by DMT1 or by ZIP8 or ZIP14 Zn transporters 56 . Our data question the DMT1 pathway because siRNA knockout of DMT1 did not affect Fe 3+ -induced Ca 2+ signalling; the detailed mechanism of Fe 3+ transport in astrocytes remains to be fully characterised.
Caveolae are specific plasmalemmal structures that form functional microdomains involved in various signalling, endocytotic and transporting events 57 . Caveolae and their main structural and regulatory proteins Caveolin-1,2,3 are present in astrocytes (with predominant expression of Cav3); astrocytic caveolae contribute to signal transduction, formation of signalling protein complexes and are involved in action of various neuroactive substances and drugs 58,59 . Caveolae are known to form functional Ca 2+ signalling units, establish links between Ca 2+ channels and various transports and may contribute to formation of plasmalemmal/ER functional domains operational in astrocytes 60 . We found that down-regulation of expression of Cav3 in cultured astrocytes substantially reduced the amplitude of Fe 2+ -evoked [Ca 2+ ] i responses. We suggest therefore that Cav3 and caveolae integrate DMT1, NKA and NCX into a single Ca 2+ signalling unit (Fig. 7); moreover exposure to iron potentiates formation of such units.
Iron homoeostasis is of fundamental importance for cells, tissues and organisms, as iron contributes to a wide range of vital biological pathways 61 . The brain contains high concentrations of bound and free iron, which participates in multiple processes from energy production to synaptic transmission 41 . Iron overload and failures in iron homoeostatic triggers neurotoxicity and is implicated in brain diseases 48 . Genetic mutations of iron regulatory proteins (the key elements of iron homoeostasis) result in iron deposition in the brain with subsequent neurodegeneration characteristic for aceruloplasminemia 62 and neuroferritinopathy 63 . In aceruloplasminemia patients, iron overload has been observed in astrocytes, particularly in the basal ganglia 64 . Ceruloplasmin oxidises Fe 2+ to Fe 3+ in order to generate the oxidised form of iron that can bind to extracellular transferrin; in astrocytes ceruloplasmin is also required for iron export 64 . Thus, in aceruloplasminemia, iron entering the CNS as ferrous might escape oxidation; hence cells exposed to the excess of Fe 2+ may readily become iron overloaded 65 . Similarly, abnormal iron accumulation has been characterised in neurodegenerative diseases including Fig. 7 Mechanisms of Fe 2+ and Fe 3+ -induced intracellular Ca 2+ signalling. Uptake of Fe 2+ is mediated by DMT1 and uptake of Fe 3+ is mediated by TFR. After entering the cytosol, Fe 2+ inhibits the activity of NKA, which causes an increase in [Na + ] i due to an unopposed influx of Na + through Na + -dependent transporters or Na + -channels. Increase in [Na + ] i switches the NCX into the reverse mode, which results in Ca 2+ influx. This influx also activates Ca 2+induced Ca 2+ release through RyR that contributes to the plateau phase of [Ca 2+ ] i response. Accumulation of Fe 2+ also promotes the formation of the functional unit of DMT1, NCX1 and NKA in the caveolae by recruiting Cav3. Uptake of Fe 3+ proceeds through Dab2-assisted internalisation; after entering the cytosol, Fe 3+ activates the PLC and stimulates InsP 3 -dependent Ca 2+ release from the ER.
Alzheimer's disease, Parkinson disease, amyotrophic lateral sclerosis and Huntington disease to name but a few 48 . Specific class of neurodegeneration with brain iron accumulation (NBIA) has been also categorised in recent years 66 .
Based on our data we propose that astrocytes mount the defence against acute iron overload. This defence includes iron accumulation through both DMT1 and TFR, redistribution of DMT1 from intracellular locations to plasmalemma and generation of Ca 2+ signals, which control astrocytic reactivity. Ironinduced Ca 2+ signalling is activated at low pathological iron concentrations (>1 μM; while physiological iron concentration in the CSF ranges between 0.3 and 0.75 μM). Importantly, two distinct signalling cascades (DMT1 Fe 2+ transport, inhibition of NKA and reversal of NCX versus Fe 3+ -TF-TFR transport, activation of PLC and generation of InsP 3 -induced Ca 2+ release) distinguish between ferric and ferrous. These distinct pathways may define very different outputs: it is known for example that activation of astroglial InsP 3 receptors type II is linked to initiation of reactive astrogliosis 67,68 . In our recent report, exposure to iron was found to stimulate the over-expression of DMT1 in astrocytes and microglia, but not in neurones, which may result in the neuroprotection by glial uptake of excessive iron 13 . Astrogliosis plays important, if not defining role in evolution of many neurological diseases 69 . Our previous experiments have shown that formation of brain deposits of iron up-regulates astroglial expression of TFR and instigates reactive astrogliosis 70 . What characterises the iron-induced reactive phenotype and what is the role of astroglial reactivity in managing excessive iron in the brain remains to be found. In conclusion, our study presented a phenomenon that iron ions (Fe 2+ and Fe 3+ ) directly induce intracellular Ca 2+ signalling and stimulate astroglial protective mechanisms against iron overload in broad pathological contexts.
Iron treatment. For preparing Fe 3+ -TF solution, ferric ammonium citrate and mouse apo-TF were incubated at a 2:1 ratio in serum-free culture medium for 1 h at 37°C 74,75 . The same concentration of apo-TF in the same volume of culture medium but without Fe 3+ was used for control treatments. For the Fe 2+ solution, FeSO 4 was freshly dissociated in serum-free culture medium at 37°C and used immediately, the same volume of serum-free culture medium was used as the control for Fe 2+ group.
RNA interfering. The cultured astrocytes were incubated in DMEM without serum for 12 hours before transfection 73,76,77 . A transfection solution containing 2 μl oligo-fectamine (Promega, Madison, WI, USA), 40 μl MEMI, and 2.5 μl siRNA (DMT1, TFR, NCX1-3, Cav-3 or Dab2) was added to the culture medium in every well for 8 h. In the siRNA-negative control cultures, transfection solution without siRNA was added. Thereafter, DMEM with three times serum was added to the cultures. These siRNA duplex chains were purchased from Santa Cruz Biotechnology (CA, USA).
Preparation of membrane caveolae. Cell homogenization and the caveolae preparation from astrocytes was made 78,79 . In brief, primary cultured astrocytes were collected and homogenised in SET (0.315 M sucrose, 20 mM Tris-HCl, and 1 mM EDTA, pH 7.4), and centrifuged for 1 h at 1,000 × g. The pellets were re-solubilised in SET and layered on Percoll (30% in SET) followed centrifugation at 1,000 × g. The pellets were re-homogenised and re-layered to three sucrose density gradient solution (80%, 30% and 5%) with ultra-centrifugation at 175,000 × g. Finally, the purified caveolae were collected and re-suspended in SET.
Co-immunoprecipitation. We used technologies of co-immunoprecipitation and subsequent western blotting to check the conjunction level between NCXs and DMT1 76 . After homogenization, protein content was determined by the Bradford method 80 using bovine serum albumin as the standard. For immunoprecipitation of NCX1-3, whole cell lysates (500 μg) were incubated with 20 μg of anti-NCX1, anti-NCX2 or anti-NCX3 antibody for overnight at 4°C. Then, 200 μl of washed protein G-agarose bead slurry was added, and the mixture was incubated for another 2 hours at 4°C. The agarose beads were washed three times with cold phosphate buffer solution (PBS) and collected by pulsed centrifugation (5 s in a microcentrifuge at 14,000 × g), the supernatant was drained off, and the beads were boiled for 5 min. Thereafter, the supernatant was collected by pulsed centrifugation, and the entire immunoprecipitates were subjected to 10% sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE).
Cytoplasm and nucleus protein extraction. The subcellular fraction was analyzed using the protein extraction Kit (P0028, Beyotime, Shanghai, China) 81 , according to the manufacturer's protocol. For further western blotting assays, cytoplasmic and nucleic protein extractions were kept under −20°C.
Monitoring of [Ca 2+ ] i . For [Ca 2+ ] i monitoring and imaging in cultured astrocytes 83,84 , after the pre-treatment with or without inhibitors or siRNA duplex chains, the primarily cultured astrocytes were loaded with 5 μM fluo-4-AM, (Thermo Fisher Scientific (Waltham, MA USA)), for 30 min. Fluo-4 signals were visualised by fluorescent microscopy (Olympus IX71, Japan). The readings from all fluo-4 positive cells in one measured field of view in each culture were included in the statistics, the fluorescence intensity of fluo-4 was normalised to the baseline intensity before stimulation. The measurements were repeated in 10 different cultures. Cultured astrocytes were also stained with specific marker sulforhodamine 101 (SR101) at 1:2000 dilution.
Two-photon in vivo Ca 2+ imaging. Adult FVB/N-Tg(GFAP-eGFP)14Mes/J transgenic mice (10-12 weeks old) were anesthetized with ketamine (80 mg/kg, i.p.) and xylazine (10 mg/kg, i.p.). Body temperature was monitored using a rectal probe, and the mice were maintained at 37°C by a heating blanket. A custom made metal plate was glued to the skull with dental acrylic cement and a cranial window was prepared over the right hemisphere at 2.5 mm lateral and 2 mm posterior to bregma. The somato-sensory cortical cells were loaded with Ca 2+ indicator Rhod-2 AM (50 μM, 1 h). The transcranial window was superfused with artificial CSF. After a stable baseline recording was obtained, Fe 2+ (100 μM) or Fe 3+ -TF (100 μM) was added for 1 min. Bandpass filters (Chroma) were 540 nm/40 nm for eGFP and 850 nm/70 nm for rhod-2 signals. Time-lapse images of astrocytic Ca 2+ signalling were recorded every 5 s using FluoView with a custom-built twophoton laser-scanning setup (Nikon AR1, Japan) 84,85 .
Intracellular Na + measurements. For monitoring intracellular ionised Na + ([Na + ] i ) in cultured astrocytes, primary cultured astrocytes were loaded with 10 μM of Na + -sensitive indicator SBFI-AM for 30 min in serum-free medium, with subsequent 1 h of washout. SBFI was alternatively excited at 340 nm and 380 nm, and the emission was monitored at 500 nm. The SBFI signals were measured by fluorescent microscopy (Olympus IX71, Japan) and expressed as a ratio (R = F 340 /F 380 ) 52 .
Immunofluorescence. The brain tissue was fixed by immersion in 4% paraformaldehyde and cut into 100 μm slices 72,82 . The cultured cells were fixed with 100% methanol at −20°C. Brain slices or cells were permeabilised by incubation for 1 h with donkey serum. Primary antibodies against DMT1 or TFR were used at a 1:100 dilution, against glial fibrillary acidic protein (GFAP) was used at 1:200 dilution. And nuclei were stained with marker 4′, 6′-diamidino-2-phenylindole (DAPI) at 1:1000 dilution. The primary antibodies were incubated overnight at 4°C and then donkey anti-mouse or anti-rabbit Cy-2/3 conjugated secondary antibody for 2 h at room temperature. Images were captured using a confocal scanning microscope (DMi8, Leica, Wetzlar, Germany).
ELISA assays. Astrocytes were incubated at 37°C in a fresh serum-free culture medium; after the treatment with Fe 2+ /Fe 3+ or inhibitors, the astrocytes were collected and centrifuged at 10,000 × g for 10 min to remove floating cells and/or cell debris at 4°C. To assay the NKA activity, a commercial ELISA kit (abx255202; Abbexa, Cambridge, UK) was used and operated as the protocols, the sensitivity is 0.19 ng/mL, the optical density (OD) was measured at 450 nm and the OD value was normalized by the control group. To assay the InsP 3 concentration 86 , the supernatant was collected and the concentration of InsP 3 assayed using a commercial ELISA kit (E-EL-0059c; Elabscience Biotechnology, Wuhan, China).
Sorting neural cells through fluorescence activated cell sorter (FACS) and quantitative PCR (qPCR). To measure the mRNA for TFR and DMT1, astrocytes expressing fluorescent marker GFP (GFAP-GFP mice) and neurons expressing fluorescent marker YFP (Thy1-YFP mice) were used; we also extracted the cerebral hemispheres tissues from wild type mice. The cells from transgenic mice were used for specific sorting of astrocytes or neurons with FACS 85,87 . The RNA of the sorted cells and cerebral tissue was extracted by Trizol. Total RNA was reverse transcribed and PCR amplification was performed in a Robo-cycler thermocycler 81,84 . The relative quantity of transcripts was assessed using five-folds serial dilutions of RT product (200 ng). RNA quantity was normalised to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and values are expressed as the ratio TFR/GAPDH or DMT1/GAPDH. Statistics and reproducibility. For statistical analysis, we used one-way analysis of variance (ANOVA) followed by a Tukey's or Dunnett's post hoc multiple comparison test for unequal replications using GraphPad Prism 5 software (GraphPad Software Inc., La Jolla, CA) and SPSS 24 software (International Business Machines Corp., NY, USA). One-way ANOVA for comparisons including more than two groups; unpaired two-tailed t-test for two-group comparisons. All statistical data in the text are presented as the mean ± SD, the value of significance was set at p < 0.05.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
Source data underlying the main and supplementary figures are available in Supplementary Data. The data that support the findings of this study are available from the corresponding author Baoman Li upon reasonable request.