Effects of iron deficiency on iron binding and internalization into acidic vacuoles in Dunaliella salina.

Uptake of iron in the halotolerant alga Dunaliella salina is mediated by a transferrin-like protein (TTf), which binds and internalizes Fe(3+) ions. Recently, we found that iron deficiency induces a large enhancement of iron binding, which is associated with accumulation of three other plasma membrane proteins that associate with TTf. In this study, we characterized the kinetic properties of iron binding and internalization and identified the site of iron internalization. Iron deficiency induces a 4-fold increase in Fe binding, but only 50% enhancement in the rate of iron uptake and also increases the affinity for iron and bicarbonate, a coligand for iron binding. These results indicate that iron deprivation leads to accumulation and modification of iron-binding sites. Iron uptake in iron-sufficient cells is preceded by an apparent time lag, resulting from prebound iron, which can be eliminated by unloading iron-binding sites. Iron is tightly bound to surface-exposed sites and hardly exchanges with medium iron. All bound iron is subsequently internalized. Accumulation of iron inhibits further iron binding and internalization. The vacuolar inhibitor bafilomycin inhibits iron uptake and internalization. Internalized iron was localized by electron microscopy within vacuolar structures that were identified as acidic vacuoles. Iron internalization is accompanied by endocytosis of surface proteins into these acidic vacuoles. A novel kinetic mechanism for iron uptake is proposed, which includes two pools of bound/compartmentalized iron separated by a rate-limiting internalization stage. The major parameter that is modulated by iron deficiency is the iron-binding capacity. We propose that excessive iron binding in iron-deficient cells serves as a temporary reservoir for iron that is subsequently internalized. This mechanism is particularly suitable for organisms that are exposed to large fluctuations in iron availability.

Iron is an essential element for survival for all living organisms, including photosynthetic organisms that have a special requirement for iron as a cofactor of multiple elements in their electron transport system. Because of its low solubility in aerobic solutions, iron is recognized as a major limiting factor for proliferation of plants and algae. To counterbalance iron limitation, photosynthetic organisms evolved efficient high-affinity iron uptake mechanisms that are induced under iron limitation. Two major mechanisms of iron acquisition studied in plants are based either on reduction of organic ferric iron chelates, via a membrane-associated ferrireductase (strategy I plants), or on release of phytosiderophores that bind ferric ions and introduce them into the cell via a special receptor (strategy II plants; Mori, 1999;Curie and Briat, 2003;Hell and Stephan, 2003). Algae appear to utilize similar mechanisms for iron acquisition. In the green algae Chlorella and Chlamydomonas and, in several diatoms, iron deficiency induces a large enhancement of ferrireductase activity associated with stimulation of highaffinity iron uptake, suggesting a redox-driven iron uptake mechanism similar to strategy I plants (Allnutt and Bonner, 1987;Eckhardt and Buckout, 1998;Lynnes et al., 1998). Other algae, such as Scenedesmus incrassatulus and Emiliania huxleyi, utilize siderophore-mediated iron uptake similar to strategy II plants and bacteria (Benderliev and Ivanova, 1994;Boye and van der Berg, 2000). However, comprehensive identification of proteins and genes associated with iron acquisition was made only in two species of algae, Chlamydomonas reinhardtii and Dunaliella salina. Several components of a copper-dependent iron uptake mechanism were identified and characterized in C. reinhardtii. They include a multicopper ferroxidase and an iron permease, suggesting that high-affinity iron uptake in C. reinhardtii is similar to that in yeast (Saccharomyces cerevisiae; Herbik et al., 2002;La Fontaine et al., 2002). A different high-affinity iron uptake mechanism was identified in the halotolerant alga D. salina, which is based on iron binding and internalization via a membrane-associated transferrin-like protein (TTf; Fisher et al., 1997Fisher et al., , 1998. In previous studies, we demonstrated that iron uptake and binding in D. salina resembles kinetic characteristics of animal transferrins and that the process consists of two distinct stages, binding and internalization of Fe 31 ions, which can be resolved by their different temperature dependence (Fisher et al., 1998;Schwarz, et al., 2003b). Iron deprivation or high salinity induces in D. salina a small stimulation of iron uptake that is correlated with the accumulation of TTf. However, the kinetic mechanism has not been investigated and the site of iron internalization was not identified. In a recent study, we found that iron dep-rivation induces in D. salina a large increase in ironbinding capacity and that this enhanced binding was correlated with accumulation of three additional plasma membrane proteins, which were found to associate with TTf: a second transferrin, termed DTf, a multicopper ferroxidase, termed DFox, and another 130-kD protein (Paz et al., 2007).
In this work, we studied the regulation of iron binding and uptake. Specifically, we characterized the kinetic changes in iron binding and uptake in response to iron deprivation or to saturation of intracellular iron stores. We show that iron deprivation or overaccumulation leads to large changes in iron binding, suggesting that this is the critical parameter in regulation of iron uptake in D. salina. We also localized the site of iron internalization within acidic vacuoles. The physiological significance of dynamic changes in iron binding is discussed.

Dunaliella Cells Contain Externally Bound Iron
The time course of iron uptake in D. salina reveals an initial time lag of 15 to 20 min. This time lag was apparent in control cells, which have been cultured with iron, but was absent in iron-deficient cells (Fig. 1A). We found that preincubation of control D. salina cells for 30 to 60 min in iron-deficient medium almost completely eliminated this apparent time lag (Fig. 1B). Conversely, preincubation of iron-deficient cells with iron introduced a similar time lag in the onset of iron uptake (Fig. 1C). These results suggest that the apparent time lag in iron uptake in D. salina may result from prebound iron.
To test directly whether D. salina cells contain prebound iron, we measured iron binding to cells before Figure 1. Prebound iron causes the time lag in Fe uptake. A, Time lag in Fe uptake in D. salina cells. Samples of 5 3 10 7 iron-deficient or ironsufficient cells were incubated with 59 Fe 31 citrate at room temperature for the indicated time points and analyzed for internalized iron content. Arrow indicates the time lag in iron-sufficient cells. B, Time lag in Fesufficient cells. Samples of 5 3 10 7 cells were either not treated (Con.) or preincubated for 1 h in the light in Fe-deficient medium, for unloading occupied iron-binding sites (PI) or for 20 min on ice with saturating Fe citrate to upload unoccupied iron-binding sites followed by washing (Fe-Cit.). After these treatments, cells were incubated with 59 Fe 31 citrate at room temperature for the indicated times and analyzed for internalized iron content. C, No time lag in Fe-deficient cells. Treatments as in B. Averages of 3 (A) or 2 (B and C) separate experiments. Cultures of iron-sufficient (1Fe; white symbols) or iron-deficient (2Fe; black symbols) cells were either not treated (circles) or preincubated (PI; squares) to unload bound Fe or preincubated with Fe citrate (Fe-Cit.; triangles) as in Figure 1. Samples were then washed and assayed for Fe binding at 4°C with 59 Fe 31 citrate for the indicated times. Representative experiment of four repetitions.
or after 1-h preincubation treatments in iron-deficient medium. Iron binding was measured at 4°C to avoid internalization of bound iron, which is strongly temperature dependent (Fisher et al., 1998). As shown in Figure 2, binding of iron to control cells was greatly enhanced by the preincubation treatment. Iron-deficient cells have a 4-fold larger iron-binding capacity, but, in contrast to iron-sufficient cells, binding was hardly affected by the preincubation treatment. Conversely, preincubation of iron-deficient cells with iron completely inhibited iron binding. These results confirm that the apparent inhibition of iron uptake results from prebound iron. Preincubation in iron-deficient medium unloads the iron-binding site (iron unloading), whereas preincubation with iron uploads the unoccupied sites (iron uploading).
In earlier studies, we have shown that the binding characteristics of iron to D. salina cells resemble transferrin iron-binding sites (Fisher et al., 1998). Moreover, we found that isolated TTf from D. salina has very similar Fe 31 -binding parameters as intact cells and that dissociation of TTf from plasma membranes completely eliminated Fe-binding activity to TTf-deficient membranes (Schwarz et al., 2003b). These studies demonstrated that most bound iron in D. salina is associated with this transferrin. To check whether Fe-deficient cells also have similar iron-binding characteristics, we compared the effects of characteristic parameters of TTf iron binding between Fe-sufficient and Fe-deficient D. salina cells (Table I). Similar dependence was found on bicarbonate, a coligand for Fe binding to transferrins, on pH, and hardly any inhibition by excess of other metal ions, suggesting high specificity for Fe 31 . These results also suggest that, in Fe-deficient cells, iron is mostly bound to transferrin.
A summary of the kinetic parameters of iron binding and uptake is presented in Table II. The initial rate of iron binding was enhanced 8-fold in iron-deficient cells as compared to control cells. This enhancement can be accounted for partly by the 4-fold increase in number of iron-binding sites and partly by the 2-fold larger calculated binding rate constant (k b ). In contrast to enhanced binding, the steady-state rate of iron uptake was enhanced by just about 50% in iron-deficient cells. The large increase in iron-binding capacity and binding-rate constant suggests that iron deprivation induces accumulation of new iron-binding sites, with slightly different kinetic properties. To test whether iron deprivation also affects the kinetic properties of iron-binding sites, we compared two kinetic parameters of iron binding between iron-deficient and ironsufficient cells, the concentration dependence on iron and on bicarbonate anions that are coligands for iron binding to transferrins and are required for binding of Fe 31 ions to D. salina transferrin, TTf (Fisher et al., 1997(Fisher et al., , 1998. As shown in Table II, iron-deficient cells display a higher affinity for both iron and bicarbonate, suggesting changes in properties of the iron-binding sites.

Tightly Bound Iron Exchanges Slowly with Medium Fe 31
The observation that D. salina cells contain prebound iron that is not removed by extensive washing is consistent with the high binding affinity of D. salina TTf for Fe 31 ions that we reported in previous studies Table I. Iron-binding characteristics in Fe-deficient and in Fe-sufficient D. salina cells Numbers in parentheses represent percent of control (no additions) activities.
Parameter/Cells 1Fe 2Fe Tf (nmol Fe/10 9 cells) 1.1 6 0.   (Fisher et al., 1998;Schwarz et al., 2003b). To find out whether iron is irreversibly bound or exchanges with medium Fe 31 ions, we added an excess of natural (nonradioactive) isotope mixture of Fe 31 , as Fe citrate, after binding 59 Fe 31 to control or iron-deficient D. salina cells. As shown in Figure 3 and Table III, there is less than 10% exchange within 1 h at 4°C. The bound iron can be stripped off effectively by EDTA and, to a lesser extent, at acidic pH. These results explain why cells that were precultured with iron do not bind 59 Fe 31 even after extensive washing because all iron-binding sites are occupied with tightly bound iron.

Bound Iron Is Committed for Internalization
In a previous study, we reported that bound iron in D. salina cells is internalized subsequent to binding in a temperature-dependent process (Fisher et al., 1998). Because in iron-deprived cells we found a large increase in number and changes in kinetic parameters of iron-binding sites, it seemed of interest to test how much of the bound iron is internalized in these cells. To test this, cells were first preincubated with 59 Fe 31 at 4°C to saturate the iron-binding sites, followed by washing and internalization in the presence of the excess natural isotope mixture Fe citrate, to exclude internalization of nonbound iron. As shown in Figure  4, practically all the bound iron was internalized in both control and iron-deficient cells. The amount of internalized iron is largely enhanced in iron-deficient cells, corresponding to the much higher binding, but the process is slower, with a t 1/2 of about 30 min compared to about 10 min in control cells. These results suggest that the massive amount of bound iron in iron-deficient cells is committed for internalization. About 30% of the bound iron in iron-deficient cells is not internalized after 1 h of incubation without iron, but can be internalized by extended incubation (data not shown). Because excess iron may be toxic to cells, we wished to test whether D. salina cells can regulate their iron-binding capacity to avoid overaccumulation of iron. To test this, cells were allowed to accumulate iron for 1.5 or 4 h and tested for ironbinding and uptake activities. To dislodge external binding sites from prebound iron, cells were exposed to two different treatments: EDTA washes, which strip off bound iron, or the 1-h unloading treatment to allow internalization of prebound iron. As is shown in Table  IV, these treatments gave different results: After 4 h of iron accumulation, EDTA stripping recovered about 90% of the original activity, whereas the unloading treatment recovered only 10% to 15%. These results suggest that, as the internal iron stores fill up, internalization is inhibited and more iron remains externally bound. This may represent a feedback mechanism to avoid iron overloading.

Bound Iron Is Internalized into Acidic Vacuoles
In mammalian cells, iron is internalized via the transferrin/transferrin receptor system into acidic vacuoles. To test whether in D. salina iron is also internalized into acidic vacuoles, we tested the effects of vacuolar inhibitors on iron uptake. In previous studies, we have identified and characterized acidic vacuoles in D. salina by staining with a fluorescent indicator quinacrine  and by localization of intravacuolar polyphosphates, which act as a pHbuffering system . We found that amines can induce alkaline stress, which is counterbalanced by massive polyphosphate hydrolysis within the vacuoles combined with amine accumulation. Thus, amines at alkaline pH can disrupt vacuolar activities in D. salina. As shown in Table V, the vacuolar H 1 -ATPase inhibitor, bafilomycin, as well as ammonia at alkaline pH, suppressed iron uptake, but only partially inhibited iron binding. Under these conditions, bafilomycin did not decrease cellular ATP nor did it inhibit uptake of phosphate (data not shown), indicating that the inhibition is specific to vacuolar  Binding of 59 Fe to iron-sufficient or iron-deficient cells was carried out for 30 min at 4°C followed by two washes in EDTA-free stop solution. Subsequently, cells were incubated for 1 h at 4°C as indicated in stop solution at pH 8 (none), at pH 8 with addition of 200 mM iron citrate (natural isotope mixture), in stop solution at pH 5 (pH 5), at pH 8 1 5 mM EDTA (EDTA), or at pH 5 1 5 mM EDTA (pH 5 1 EDTA). Cells were washed and tested for bound 59 Fe.  (Parmley et al., 1978;Kim et al., 2003;Vasconcelos et al., 2003;Arbab et al., 2006;Tallheden et al., 2006) and monitored stained cells by electron microscopy. Bound iron particles were clearly visualized at the extracellullar cell surface (Fig. 5A, inset). The remarkable difference before and after internalization was the dense iron staining within vacuolar-like structures (compare Fig.  5, A and B). At high magnification, iron within some of the vacuoles appears to be associated with a delicate membrane-like reticulum (Fig. 5, E and F) and, in others, is aggregated in dense plaques (Fig. 5, C and D). This stained iron particles resemble in size and general appearance early ultrastructural localizations of iron within phagosomes of macrophage cells (Parmley et al., 1978). The significance of the different patterns of vacuolar iron staining is not clear. They may represent different stages in vacuolar maturation after endocytosis; possibly, the reticular structures may represent residual plasma membrane structures in immature vacuoles, shortly after endocytosis, which later mature to yield aggregated iron morphology. To further characterize these iron-accumulating structures, we treated D. salina cells with the fluorescent indicator LysoSensor Green, which specifically stains acidic vacuoles. As shown in Figure 6A, the stain accumulates in vacuolar structures similar in size and localization to the iron-accumulating vacuoles. To further characterize these vacuoles, we immunolabeled D. salina cells with polyclonal antibodies against a vacuolar marker previously reported to label acidic vacuoles in the green alga C. reinhardtii, the H 1 -transporting pyrophosphatase (Ruiz et al., 2001). Antibodies reacted with a single component on western immunoanalysis corresponding to its predicted M r (data not shown). As shown in Figure 6B, anti H 1 -pyrophosphatase labeled structures similar in size and localization to acidic vacuoles. Unfortunately, it is not possible to carry out colocalization of the iron stain with either immunolocalization or LysoSensor Green staining because of the different required preparation procedures (acid treatment, fixation, and detergent permeabilization or vital stain, respectively). To test whether iron internalization indeed results from endocytosis of iron-binding proteins, we tagged membrane surface proteins in irondeficient cells with a membrane-impermeable biotin derivative and monitored biotin-tagged proteins labeled with the fluorescein avidin with a confocal microscope. We found that, in iron-deficient cells, biotin tags almost exclusively three protein bands that were identified as the major iron deprivation-induced plasma membrane proteins (Paz et al., 2007): two transferrin-like proteins, TTf (Fisher et al., 1997(Fisher et al., , 1998 and DTf (Schwarz et al., 2003a), and a 130-kD band that was found to contain two proteins, a multicopper ferroxidase, termed DFox, and a second unknown protein. In stained cells that were kept at 4°C, biotin-tagged label was confined to the outer cell surface (Fig. 6C). However, after internalization of bound iron for 1 h at 24°C, tagged vacuolar-like structures appeared at the apical side of the cells (Fig. 6D). Preincubation with the vacuolar inhibitor bafilomycin prevented the appearance of biotin-tagged vacuoles, but induced structural deformation that appears like thickening of the membrane surface at the apical side of the cells (Fig. 6E). Taken together, these results suggest that iron is internalized into acidic vacuoles by  Table IV. Preaccumulation of iron inhibits iron binding and uptake Accumulation of iron was performed by incubation of control (1Fe) or iron-deficient (2Fe) cells for 0 to 4 h in the light with 10 mM iron citrate (natural isotope mixture). Cells were then treated either by preincubation for 1 h in iron-deficient medium (Inc. 2Fe) or washed twice in iron uptake stop solution containing 5 mM EDTA (1EDTA) and twice in EDTA-free solution. All cell samples were analyzed for iron binding and uptake activities. endocytosis of iron-binding proteins and that the process is inhibited by bafilomycin.

DISCUSSION
As shown in our previous work (Fisher et al., 1998) and here, D. salina binds large amounts of Fe 31 ions. Iron-deficient cells can bind about 4 nmol Fe/10 9 cells or 2.5 3 10 6 Fe 31 ions/cell, yielding an exceptionally high calculated density of about 10 5 Fe/mm 2 (average cell diameter of 5 mm). Iron can be accepted from a variety of Fe 31 ligands, including citrate, EDTA, or desferal in the presence of bicarbonate (Pick, 2004) and is so tightly bound that it cannot be washed away or exchanged by excess medium iron and requires acidification and EDTA for effective release. In fact, the large increase in iron binding enables D. salina to bind in a short time sufficient iron to satisfy its whole iron budget and then slowly internalize the bound iron. The tight binding of iron, which is followed by energydependent internalization (Fisher et al., 1998), renders this process practically irreversible. These properties are characteristic of iron binding to transferrins, consistent with our previous demonstration that D. salina acquires iron via a surface-exposed transferrin, TTf.
The simplest kinetic model to describe iron uptake in D. salina according to our results is the following (see Supplemental Model S1): Where Fe 31 represents free ferric ions, Tf-iron-binding proteins [Tf-Fe] bound iron and [Fe-V] internalized iron. k b and k int represent the rate constants for iron binding and internalization, respectively. k b is much  Arrows, Iron-loaded vacuoles. Scale, 1 mm in A and B; 0.5 mm in A (inset) and C to F. larger than k int ; therefore, the overall rate of Fe uptake is dominated by the rate of internalization, except at limiting Fe concentration. Overaccumulation of iron, according to this model, is predicted to inhibit k int , leading to occupation of all iron-binding sites (accumulation of [Fe-Tf]) and to inhibition of Fe uptake. Iron deprivation, as shown, increases primarily Tf (4-fold) and k b (2-fold), but not k int or [Fe-V], as manifested by the large increase in Fe-binding capacity, Fe-binding rate constant, and the modest increase in Fe uptake, respectively. The 4-fold increase in Fe-binding capacity indicates induction of more iron-binding proteins at the plasma membrane. The increase in Fe-binding rate constant and the lower K m for iron and for bicarbonate indicate intrinsic changes in the iron-binding sites.
Conversely, the small increase in rate of internalization suggests that the number of Fe-binding units, [Fe-Tf], does not increase very much. Our kinetic results can be accommodated with the above kinetic model by inferring that, in Fe-deficient plasma membranes, small ironbinding units made of Fe-Tf are converted to larger iron-binding complexes with higher iron-binding capacity, without major changes in the number of Febinding units (or [Fe n -Tf] x / [Fe m -Tf] y , where m $ n and x 5 y [approximately]). As mentioned above, we found recently that the increase in Fe binding in Fe-deficient cells is correlated with induction of three plasma membrane proteins, DTf (Schwarz et al., 2003a), DFox, and p130B, which associate with TTf (Paz et al., 2007). These results are consistent with the proposed kinetic model. The reasons why Dunaliella selected transferrins for iron binding are not clear. One reason may be that iron binding to transferrins is quite resistant to high salt. Another reason may be the high affinity and selectivity of transferrins for ferric ions, which would exclude competition by common divalent and trivalent metal ions that are often abundant in hypersaline and brackish water solutions. The kinetic response to Fe deprivation differs very much from plants, other algae, yeast, and bacteria, where iron deprivation is usually associated with a dramatic increase in the rate of iron uptake.
However, such a mechanism is suitable for an organism that is exposed to large periodic changes in iron availability. The natural living habitats of Dunaliella are saline lakes and the open sea, which are often poor in bioavailable iron and obtain their iron supply from winter floods or dust carried by storms. The proposed mechanism of iron acquisition in D. salina offers a competitive advantage in comparison to carriermediated uptake mechanisms under such conditions because it enables efficient capture within a short time of large quantities of iron from diverse iron-binding ligands. Possible drawbacks of massive accumulation of proteins at the plasma membrane surface are the heavy energetic cost of protein synthesis, and it may interfere with other membrane functions because it occupies a large area of the plasma membrane. Cells were preincubated for 10 min with the fluorescent probe. Acidic vacuoles are stained in green (arrow) and chlorophyll, which marks the cup-shaped chloroplast, is stained in red. B, Immunolocalization of acidic vacuoles with antibodies against V-H 1 -PPase. Cells were fixed, permeabilized, and incubated for 12 h with antibodies against C. reinhardtii V-H 1 -PPase and next with fluorescein goat anti-rabbit antibodies. Fluorescein indicating V-H 1 -PPase is stained in green (arrow) and chlorophyll is stained in red. C to E, Biotin tagging of surface proteins. Cells were labeled with biotin, washed, and either fixed immediately (C) or treated to bind and internalize iron in the absence (D) or presence (E) of 5 mM bafilomycin. Fixed cells were permeabilized and incubated with fluorescein avidin. Fluorescein, representing surface labeled proteins, is stained in green (arrow) and chlorophyll is stained in red.

Iron Binding and Internalization in Dunaliella
Plant Physiol. Vol. 144, 2007 The finding that D. salina internalizes iron into acidic vacuoles resembles iron uptake via the transferrin/ transferrin receptor system in mammalian cells, but is unusual for plants that store iron primarily in the chloroplast bound to ferritin (Curie and Briat, 2003). Nevertheless, plant seeds store iron in acidic vacuoles (Lanquar et al., 2005) and it was recently reported that, in the presence of sufficient phosphate, Arabidopsis (Arabidopsis thaliana) stores iron as phosphate-iron complexes within the vacuole (Hirsch et al., 2006). Dunaliella, as well as other algae, store phosphate primarily as insoluble polyphosphates within acidic vacuoles Ruiz et al., 2001). Vacuolar polyphosphates may serve as an excellent sink for iron storage by providing a high-capacity iron-binding reservoir. Thus, acidic vacuoles may serve a central role in the regulation and storage of iron in D. salina. One manifestation of vacuolar regulation is the inhibition of iron binding and uptake as the vacuoles fill up. It should be realized, however, that the vacuoles are not the final station of iron in Dunaliella. We estimated from 59 Fe subcellular fractionation that, under steadystate growth conditions in Fe-sufficient medium, 70% to 80% of the cellular iron in D. salina is localized within the chloroplast, partly stored in ferritin, and partly incorporated into different proteins (U. Pick, unpublished data). Thus, similar to plants, in Dunaliella the chloroplast also seems to be the major consumer and storage site for iron and the acidic vacuoles are just an intermediary storage station. However, for an organism that takes up in a relatively short time enormous amounts of iron, vacuoles can serve as a safe buffering checkpoint that enables control of cytoplasmic levels of iron and avoidance of iron toxicity. Thus, Dunaliella has evolved a very exceptional mechanism for iron acquisition and homeostasis that may have special advantages for adaptation to hypersaline and iron-deficient environments.

Algal Strain and Growth Condition
Dunaliella salina, a green species, was obtained from the culture collection of Dr. W.H. Thomas (La Jolla, CA). All culture glassware was washed in acid and thoroughly rinsed with Mili-Q water. Cells were cultured in Fe-sufficient or in Fe-deficient medium as previously described (Fisher et al., 1997(Fisher et al., , 1998Schwarz et al., 2003a). Standard growth medium contained 1 M NaCl, with or without FeCl 3 /EDTA (1.5/6.0 mM) unless otherwise indicated.

Prussian Blue Staining of Iron
Ultrastructural localization of iron was made with acid ferrocyanide staining (Parmley et al., 1978). Iron-deficient cells were collected by centrifugation, washed once in 1 M NaCl, and suspended in iron uptake buffer containing growth medium without iron, 50 mM HEPES, pH 8.0, and 5 mM NaHCO 3 . Cells were incubated with 10 mM Fe citrate for 30 min on ice (bound iron) or for 60 min at room temperature in the light under continuous shaking (internalized iron). Cells were fixed in 3.7% paraformaldehyde (p-formaldehyde) plus 1% glutaraldehyde and were next reacted for 30 min with fresh Perl's ferrocyanide solution, consisting of 0.5 g of potassium ferrocyanide in 49.5 mL of distilled water to which 0.5 mL of concentrated HCl (32%) was added. After extensive washing, fixed samples were rehydrated and embedded in Epon. Thin sections were examined in a Tecnai T12 (FEI) electron microscope operating at 120 kV and images were recorded using a MegaView III CCD camera (SIS GmbH).

Staining Acidic Vacuoles with LysoSensor Green
D. salina cells (2 3 10 7 cells/mL in fresh growth medium, pH 8) were incubated for 10 min with 2 mM LysoSensor Green DND-189 (Molecular Probes) from a 1 mM stock solution in dimethylsulfoxide, washed twice, incubated for 10 min with 0.25% p-formaldehyde, and placed on a polylysinecoated glass slide. Cells were viewed and photographed in a laser-scanning confocal microscope (Fluoview FV500; Olympus) using an excitation wavelength of 458 nm and emission of 510 to 550 nm or of .660 nm to view LysoSensor Green and chlorophyll, respectively.

Immunolocalization
To immunolocalize acidic vacuoles, D. salina cells were labeled with antibodies against Chlamydomonas reinhardtii vacuolar H 1 -pyrophosphatase (V-H 1 -PPase), obtained from Dr. R. Decampo (University of Illinois, Urbana, IL). Cells were fixed for 1 h with freshly prepared 3.7% p-formaldehyde, permeabilized for 5 min with 1% NP-40, washed twice, incubated for 60 min with blocking medium containing 4% goat serum, 1% bovine serum albumin, and 0.05% Tween 20 in phosphate-buffered saline. The preparation was incubated for 12 h with rabbit anti V-H 1 -PPase serum (at 1:200 dilution) in blocking solution at 5°C, washed twice, incubated with fluorescein goat antirabbit polyclonal antibodies (Jackson ImmunoResearch Laboratories), washed twice, and viewed and photographed in a confocal microscope as above at excitation wavelength 488 nm and emission wavelengths 505 to 525 nm or .660 nm to view fluorescein and chlorophyll, respectively.

Iron Uptake Assay
Iron uptake in D. salina cells was measured as previously described (Fisher et al., 1998). Cells (5 3 10 7 cells/mL) were incubated with 59 Fe citrate in the light for 1 h, the uptake was stopped by dilution into an acidified EDTA medium (pH 5.5), left for 15 to 30 min on ice, collected by centrifugation, and washed once again in the same medium. This treatment was found to remove all externally bound iron, but not internalized iron. Washed cells were counted in a b-counter.

Iron-Binding Assay
Before the binding assay, cells were treated to unload prebound iron by preincubation for 1 h at 23°C in buffer containing growth medium without iron, 50 mM HEPES, pH 8.0, and 5 mM NaHCO 3 in the light. Iron binding to intact cells was measured essentially as previously described (Fisher et al., 1998). In brief, cells were incubated for 30 min at 4°C with 59 Fe citrate, washed in EDTA-free buffer (stop solution), and counted in a b-counter. Under these conditions, .95% of the cell-associated iron was found to be externally bound to transferrins (bicarbonate dependent, EDTA sensitive). Plasma membrane vesicles were incubated for 1 h on ice with 59 Fe citrate in 50 mM Na-HEPES 8.0, 0.2 M KCl, and 5 mM NaHCO 3 and passed through semidry Sephadex G-50 columns to eliminate unbound 59 Fe as previously described (Schwarz et al., 2003b).

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Model S1. Kinetic model of Fe binding and internalization.

ACKNOWLEDGMENT
We wish to thank Dr. Roberto Docampo from the University of Illinois for providing antibodies against C. reinhardtii V-H 1 -PPase.