Exosome release is regulated by a calcium-dependent mechanism in K562 cells.

Multivesicular bodies (MVBs) are endocytic structures that contain small vesicles formed by the budding of an endosomal membrane into the lumen of the compartment. Fusion of MVBs with the plasma membrane results in secretion of the small internal vesicles termed exosomes. K562 cells are a hematopoietic cell line that releases exosomes. The application of monensin (MON) generated large MVBs that were labeled with a fluorescent lipid. Exosome release was markedly enhanced by MON treatment, a Na+/H+ exchanger that induces changes in intracellular calcium (Ca2+). To explore the possibility that the effect of MON on exosome release was caused via an increase in Ca2+, we have used a calcium ionophore and a chelator of intracellular Ca2+. Our results indicate that increasing intracellular Ca2+ stimulates exosome secretion. Furthermore, MON-stimulated exosome release was completely eliminated by 1,2-bis(2-aminophenoxy)ethane-N,N,N',N'-tetraacetic acid acetoxymethyl ester (BAPTA-AM), implying a requirement for Ca2+ in this process. We have observed that the large MVBs generated in the presence of MON accumulated Ca2+ as determined by labeling with Fluo3-AM, suggesting that intralumenal Ca2+ might play a critical role in the secretory process. Interestingly, our results indicate that transferrin (Tf) stimulated exosome release in a Ca2+-dependent manner, suggesting that Tf might be a physiological stimulus for exosome release in K562 cells.

Multivesicular bodies (MVBs) 1 are endocytic organelles that contain small internal vesicles generated from inward budding of the limiting membrane. In antigen-presenting cells, the fusion of these MVBs with the plasma membrane leads to the release of internal vesicles into the extracellular space (1). The released vesicles, termed exosomes (for a review see Refs. 2 and 3), were initially described in reticulocyte maturation, where their function was to discard plasma membrane proteins that were no longer necessary, such as the transferrin receptor (4 -6). Although other plasma membrane proteins (e.g. acetylcholinesterase) are secreted via exosomes, these small vesicles are devoid of both cytosolic proteins and proteins associated with other intracellular organelles, indicating that only a select group of macromolecules is shed via this pathway. Exosomes are also secreted by other cell types such as activated platelets, which may function in signaling/adhesion, thus having a role at sites of vascular injury (7,8). Exosomes from cytotoxic T cells and B lymphocytes may be involved in targeting molecules for cell death (9) or antigen presentation (10,11).
Despite the diverse extracellular functions that are carried out by exosomes, very little is known about the molecular machinery involved in either the formation of the MVBs or in the exosome secretory process. We have recently shown that in K562 cells, a human erythroleukemia cell line, overexpression of Rab11 regulates the exosome pathway (12). Interestingly, treatment of green fluorescent protein-Rab11-transfected cells with the ionophore monensin (MON) generated large MVBs decorated with Rab11 and labeled with a fluorescent lipid that accumulates in exosomes. MON, a membrane-permeable Na ϩ ionophore that mediates an antiporter activity exchanging Na ϩ ions with H ϩ ions (13), acts on acidic intracellular organelles such as endosomes and lysosomes, causing swelling of these vesicles. MON is also known to induce Ca 2ϩ entry by reversed activity of the Na ϩ /Ca 2ϩ exchanger (14 -16).
A rise in intracellular Ca 2ϩ concentration, a universal intracellular signal (for a review see Refs. 17 and 18), is necessary to induce regulated secretion in most cell types (reviewed in Refs. 19 and 20). During regulated exocytosis, the membrane of a secretory vesicle fuses with the plasma membrane in a tightly controlled Ca 2ϩ -triggered reaction. In endocrine cells, secretory granules contain large amounts of Ca 2ϩ ions, and it has been suggested that the high intragranular Ca 2ϩ concentration is needed to sustain optimal exocytosis (21). Because MON generates large MVBs in K562 cells, the aim of the present study was to determine whether MON affects exosome release and establish whether Ca 2ϩ is involved in this process. Our results indicate that both MON treatment and a rise in intracellular Ca 2ϩ markedly stimulate exosome secretion. Furthermore, the MON-stimulated exosome release was a Ca 2ϩ -dependent process. Interestingly, we have also observed that MON induced the accumulation of Ca 2ϩ in the enlarged MVBs, suggesting that intravesicular Ca 2ϩ might be involved in the secretory step. To determine whether a physiological signal might regu-late the Ca 2ϩ -dependent exosome release, cells were incubated with transferrin (Tf). Our results indicate that Tf stimulates exosome release in a Ca 2ϩ -dependent manner.
Exosome Isolation-Exosomes were collected from 10 ml of K562 media (15-20 ϫ 10 6 cells) cultured over 7-15 h. The culture media were collected on ice, centrifuged at 800 ϫ g for 10 min to sediment the cells, and then centrifuged at 12,000 ϫ g for 30 min to remove the cellular debris. Exosomes were separated from the supernatant by centrifugation at 100,000 ϫ g for 2 h. The exosome pellet was washed once in a large volume of PBS and resuspended in 100 l of PBS (exosome fraction).
Quantitation of Released Exosomes-The amount of released exosomes was quantitated by measuring the activity of acetylcholinesterase, an enzyme that is specifically directed to these vesicles (12). Acetylcholinesterase activity was assayed following a previously described procedure (22). Briefly, 25 l of the exosome fraction were suspended in 100 l of phosphate buffer and incubated with 1.25 mM acetylthiocholine and 0.1 mM 5,5Ј-dithiobis(2-nitrobenzoic acid) in a final volume of 1 ml. The incubation was carried out in cuvettes at 37°C, and the change in absorbance at 412 nm was followed continuously. The data represent the enzymatic activity at 20 min of incubation.
As an independent assay, exosomes were quantitated by determining the levels of the protein Hsc70 by Western blot. Samples of the exosomal fraction (15 l) were solubilized in reducing SDS loading buffer, incubated for 5 min at 95°C, run on 10% polyacrylamide gels, and transferred to an Immobilon (Millipore) membrane. The membranes were blocked for 1h in Blotto (5% nonfat milk, 0.1% Tween 20, and PBS) and subsequently washed twice with PBS with 0.1% Tween 20 or Tris-buffered saline with 0.1% Tween 20. Membranes were incubated with primary antibodies and peroxidase-conjugated secondary antibodies. The corresponding bands were detected using an enhanced chemiluminescence detection kit (Pierce) and quantitated by densitometry.
Labeling MVBs with the Fluorescent Lipid N-Rh-PE and Fluo3-AM for Imaging Calcium-The fluorescent phospholipid analog N-Rh-PE was inserted into the plasma membrane as described previously (23). Briefly, an appropriate amount of the lipid, stored in chloroform/methanol (2:1), was dried under nitrogen and subsequently solubilized in absolute ethanol. This ethanolic solution was injected with a Hamilton syringe into serum-free RPMI (Ͻ1%, v/v) while vigorously vortexing. The mixture was then added to the cells, which were incubated for 60 min at 4°C. After this incubation period, the medium was removed, and the cells were extensively washed with cold PBS to remove excess unbound lipids. After the addition of complete RPMI medium and Fluo3-AM (15 M), labeled cells were cultured for 2-3 h under conditions as described and washed twice with ice-cold PBS. Cells were mounted on coverslips and immediately analyzed by fluorescence microscopy. In some experiments, the cells were preloaded with Fluo3-AM by incubating for 60 min at 37°C before labeling with the fluorescent lipid. No major differences were observed between these experimental procedures.
Fluorescence Microscopy-K562 cells were analyzed using an inverted microscope (Nikon Eclipse TE 300, Japan) equipped with the following filter systems: excitation filter 450 -490 nm, barrier filter 515 nm to visualize Fluo3-AM; and excitation filter 510 -560 nm, barrier filter 590 nm to localize N-Rh-PE. Images were captured with a CCD camera (Orca I, Hamamatsu) and processed using the program Meta-Morph 4.5 (Universal Images Corporation). Some images were obtained with a Nikon Confocal C1 and processed with the EZ-C1 program.
Measurement of Intracellular Calcium Concentration-Cells were incubated in the presence of 10 M Fura2-AM for 60 min at 37°C. They were washed to remove the extracellular dye and resuspended in com-plete RPMI medium containing 1 ϫ 10 6 cells/ml. Fura2-AM loaded cells were protected from light. Experiments were completed within 2 h. Changes in fluorescence after the addition of 7 M MON or by adding 30 M BAPTA-AM before MON were analyzed in a Hitachi F-2000 fluorescence spectrophotometer.

Monensin Induces the Formation of Large MVBs and Stim-
ulates Exosome Secretion-K562 cells are human erythroleukemic cells that secrete exosomes (24), the small internal vesicles released into the extracellular media by fusion of MVBs with the plasma membrane (PM). It has been shown by electron microscopy that treatment of K562 cells with the ionophore MON causes the formation of dilated MVBs (25,12). However, the mechanism by which these large MVBs are formed and the effect of MON on exosome release have not been explored. To get insights into these issues, MVBs in K562 cells were labeled with the fluorescent lipid analog N-Rh-PE. Sucrose gradient analysis and immunoisolation experiments have demonstrated that this lipid is efficiently internalized via endocytosis and targeted to the MVBs. Indeed, N-Rh-PE accumulates in exosomes that are eventually secreted into the extracellular medium (12,26). The lipid N-Rh-PE was first bound to the PM at 4°C, and cells were washed and subsequently incu- Because fusion of the MVBs with the PM results in the release of exosomes, we tested the effect of MON on the release of exosomes from K562 cells. Exosomes are enriched in proteins such as the transferrin receptor (TfR), Hsc70, and acetylcholinesterase (AChE) (27). Therefore, exosomes were quantitated in the exosomal fraction by measuring the activity of AChE (see "Experimental Procedures"). Also, the amount of Hsc70 and TfR was determined by Western blot as described previously (12). Exosomes were harvested from the extracellular media after 7-h incubations with different concentrations of MON and quantitated by determining the levels of the proteins Hsc70 ( Fig. 1C) and TfR (not shown) by Western blot. As shown in Fig.  1C, MON induced a marked increase in exosome release in a concentration-dependent manner. A similar increase was observed by measuring, in the exosomal fraction, the activity of AChE (Fig. 1D), which was maximal at 10 M MON. At higher MON concentrations some alterations in cell viability were observed as assessed by trypan blue exclusion, for which reason a 7 M concentration of MON was used in the rest of the experiments. At this concentration, cells were also assayed for apoptosis by staining the nucleus with Hoechst 33342 (Molecular Probes). No morphological evidence of apoptotic nuclei was observed (data not shown).
In some experiments, the amount of exosomes released was also quantified by assaying the fluorescent lipid N-Rh-PE. As mentioned above, this lipid accumulates in intracellular vesicles that are ultimately secreted into the extracellular medium as exosomes. As expected, MON also increased the release of exosomes labeled with the fluorescent lipid (data not shown). Taken together the results indicate that MON not only generates large MVBs but also increases the secretion of the internal vesicles termed exosomes.
A Calcium-dependent Mechanism Is Involved in the Monensin-stimulated Exosome Release-It has been shown that MON, a Na ϩ ionophore, can increase cytosolic Ca 2ϩ by reversing the Na ϩ /Ca 2ϩ exchange mechanism (14 -16). Therefore, to assess whether in our system the enhanced exosome release induced by MON was due to an increase in intracellular Ca 2ϩ , we first measured whether MON could modify the intracellular Ca 2ϩ concentration in K562 cells. For this purpose, cells were loaded for 1 h at 37°C with 10 M Fura-2/AM. Subsequently, the intracellular Ca 2ϩ concentration was measured by spectrofluorometry for different periods of time after the addition of MON. Fig. 2A shows that there was an initial Ca 2ϩ peak and a subsequent marked rise in intracellular Ca 2ϩ that was sustained over the 2-h period tested (Fig. 2B). The MON-induced Ca 2ϩ rise was abolished by the previous addition of the intracellular Ca 2ϩ chelator BAPTA-AM ( Fig. 2A). Interestingly, in the presence of the extracellular Ca 2ϩ chelator EGTA, MON induced the initial rise, which was likely due to Ca 2ϩ release from intracellular stores. However, no sustained increase was observed, indicating that the latter is a result of Ca 2ϩ influx from the extracellular environment (data not shown).
The results suggest that the increase in exosome release might be due to a Ca 2ϩ -dependent mechanism. To test this hypothesis, we assessed whether the MON effect on exosome release could also be prevented by Ca 2ϩ chelators. To chelate the Ca 2ϩ present in the extracellular media, cells were incubated for several hours in the presence of 1.5 mM EGTA. Under these conditions, the free Ca 2ϩ concentration was less than 10 nM as calculated with the Sliders program (see "Experimental Procedures"). BAPTA-AM was used to chelate intracellular Ca 2ϩ , because this is a membrane-permeable agent that efficiently chelates Ca 2ϩ . The released exosomes were collected from the media and quantitated by measuring AChE activity as indicated under "Experimental Procedures." As shown in Fig. 2C, both EGTA and BAPTA-AM decreased, although slightly, the basal release of exosomes. Moreover, the MON-dependent increase was completely abrogated by the Ca 2ϩ chelators, and no additive effects were observed when both chelators were added together (data not shown). The result clearly indicates that Ca 2ϩ from the extracellular media and also from intracellular stores is required for the MON-induced exosome secretion.
Ca 2ϩ involvement in exosome release was evaluated using the Ca 2ϩ ionophore A23187. As shown in Fig. 3A, incubation with the Ca 2ϩ ionophore stimulated exosome secretion to a similar extent as MON. No additive effects were observed when both agents were added together. As expected, the secretory effect of the Ca 2ϩ ionophore was inhibited by the chelators EGTA or BAPTA-AM (Fig. 3B).
It is known that MON acts on acidic compartments by altering the proton gradient across vesicle membranes, resulting in a Ca 2ϩ movement into the cytosol (28). We tested the effect of two agents known to alter the pH of vacuolar compartments, the weak base chloroquine and the vacuolar proton pump inhibitor bafilomycin A1 (29). Previous work has shown that these compounds may also discharge intracellular Ca 2ϩ pools from acidic compartments (30,31). We have evidence that chloroquine elevated intracellular Ca 2ϩ in a similar manner to MON (not shown). As shown in Fig. 4A, chloroquine stimulated the release of exosomes, although to a lesser degree than MON. Non-additive effects were observed by the addition of chloroquine together with MON. Bafilomycin also increased the release of exosomes (Fig. 4B). Both chloroquine and bafilomycin effects were abrogated by clamping extracellular Ca 2ϩ with the chelator EGTA, indicating that these compounds indeed act via a calcium-dependent mechanism.
Visualizing a MVB Calcium Pool by Fluo3-AM Imaging-Numerous reports indicate the existence of several intracellular Ca 2ϩ pools (32); for a review see Refs. 33 and 34. Fluo3-AM is a membrane-permeant compound that accumulates in the cytoplasm where cytosolic esterases clip the AM groups, rendering the fluorescent probe membrane impermeable. However, it has been shown that, when used at higher concentrations, part of this indicator is capable of accumulating also in intracellular compartments and can be used as an indicator for intracellular Ca 2ϩ stores (35). Cells were incubated with Fluo3-AM for 1 h at 37°C to visualize the calcium-containing compartments. MVBs were labeled with the fluorescent lipid N-Rh-PE as mentioned above, and the cells were subsequently incubated with the indicated agents for 3 h at 37°C. As shown in Fig. 5, the large MVBs induced by MON treatment were clearly labeled by Fluo3-AM, indicating that Ca 2ϩ accumulates in these intracellular compartments. Similarly, Ca 2ϩ was also present in the large MVBs formed by chloroquine treatment. The presence of BAPTA-AM depleted the MVBs calcium pool in both conditions. Strikingly, the size of the MVBs was markedly reduced, indicating that a calcium-dependent mechanism is involved in the development of the gigantic MVBs formed by MON or chloroquine treatment.

Calcium Levels Regulated by IP 3 Receptors and a Thapsigargin-sensitive Ca 2ϩ
Pump Are Involved in the Release of Exosome-It is well established that thapsigargin (TG) causes a rapid inhibition of the calcium-ATPase pump present in the membranes of the endoplasmic reticulum (36), followed by a fast Ca 2ϩ leak from other Ca 2ϩ stores as well as influx from the extracellular media. This leads to a rapid and pronounced increase in the concentration of cytosolic-free calcium. As expected, treatment of K562 wells with this inhibitor stimulated exosome secretion in a similar manner as MON, and this effect was also blocked by EGTA (Fig. 6A). These findings confirmed a role for Ca 2ϩ in the exosome secretory pathway and the participation of a TG-sensitive Ca 2ϩ pump in the process.
Because TG stimulated exosome release, we were interested in knowing whether the large MVBs developed by MON were also formed by treatment with TG. As shown in Fig. 6B, TG neither generated large MVBs nor impaired the formation of the MON-induced gigantic structures that were filled with calcium. This suggests that an increase in cytosolic Ca 2ϩ is not by itself enough to generate the enlarged MVBs, despite being sufficient to stimulate exosome secretion.
The phosphoinositide signaling cascade plays a prominent role in the mobilization of Ca 2ϩ from intracellular stores (for a review see Refs. 37 and 38). The receptors for the second messenger, inositol 1,4,5-trisphosphate (IP 3 ), constitute a family of Ca 2ϩ channels responsible for the mobilization of intracellular Ca 2ϩ stores. The increase in the levels of IP 3 result in the opening of Ca 2ϩ channels present in the endoplasmic reticulum and the subsequent release of Ca 2ϩ into the cytosol (39). To test whether this type of channel might be involved in the calcium-induced exosome release, cells were incubated with 2-APB, a membrane-permeable inhibitor of IP 3 -induced Ca 2ϩ release. Fig. 6C shows that 2-APB inhibited monensin-stimulated exosome release. Similar results were obtained with xestopongin, a potent blocker of IP 3 receptors (data not shown), indicating that a transient Ca 2ϩ rise mediated by the stimulation of an IP 3 receptor is critical for the MON-stimulated exosome release. However, the formation of the large Ca 2ϩ -rich MVBs induced by MON was not completely abrogated by 2-APB (Fig. 6D), although there was a decrease in the size of the MVBs compared with the vesicles generated by MON in the absence of 2-APB. Vesicle area in the MON-treated cells was 436 Ϯ 30 (relative units), whereas the addition of 2-APB reduced the size to 182 Ϯ 16 (n ϭ 50 vesicles counted). This suggests that the release of Ca 2ϩ via the IP 3 -sensitive Ca 2ϩ channels contributes, at least in part, to the generation of the enlarged MVBs.

The Formation of the Gigantic MVBs Filled with Calcium
Are Completely Blocked by Amiloride-The results presented here indicate that Ca 2ϩ is absolutely required for generation of the enlarged MVBs, because these structures are not formed in the presence of BAPTA-AM (see above). However, even though IP 3 -sensitive Ca 2ϩ channels seem to participate in the process, the development of the large MVBs was only partly decreased by specific modulators. This implies that another type of Ca 2ϩ channel (see "Discussion") is involved in the MON-dependent Ca 2ϩ rise and the accumulation of Ca 2ϩ in the MVBs.
It is known that the rapid sodium influx initiated by MON increases cytosolic Ca 2ϩ by reversing the Na ϩ /Ca 2ϩ exchange mechanism. Therefore, because the activity of a Na ϩ /Ca 2ϩ exchanger seems to be critical for the MON effect, we tested dimethyl amiloride, an inhibitor of the H ϩ /Na ϩ and Na ϩ /Ca 2ϩ exchangers. As shown in Fig. 7A, amiloride decreased the basal exosome release and also completely inhibited the MON-stimulated secretion of exosomes. As expected, the formation of the gigantic Ca 2ϩ -filled MVBs generated by MON-treatment was completely abrogated by amiloride. In contrast, the process was not affected by verapamil, an inhibitor of a voltage-dependent Ca 2ϩ channel (data not shown). These results are consistent with the idea that activation of a Na ϩ /Ca 2ϩ exchanger is a prerequisite for the Ca 2ϩ rise in the cytoplasm that leads to exosome secretion and the formation of the enlarged MVBs filled with Ca 2ϩ generated by MON.
Transferrin Stimulates Exosome Release in a Calcium-dependent Manner-Taken together the results discussed above clearly indicate that Ca 2ϩ is a key participant in the exosome release process. Therefore, we were interested in addressing whether a physiological stimulus might also enhance exosome secretion in a Ca 2ϩ -dependent manner. As mentioned previ- ously, K562 is a human erythroleukemia cell line that presents high levels of TfR, and, because it has been shown that binding of Tf to its receptor increases intracellular Ca 2ϩ concentration (40), Tf might be a candidate for regulating exosome secretion. Therefore, we first assessed whether Tf increases intracellular Ca 2ϩ in K562 cells. For this purpose, cells were loaded with Fura-2/AM as described above, and 20 g/ml human Tf was added to the incubation media. As described previously for other cell types (40), Tf induced an increase in Ca 2ϩ that was sustained for the whole 60-min period studied (Fig. 8A).
We next addressed whether adding Tf to the culture media modifies exosome secretion. Cells were incubated for 12 h in the presence of 20 g/ml human Tf, and exosomes were collected from the incubation media as described above. As shown in Fig.  8B, Tf stimulated exosome release, an effect that was hampered by EGTA or BAPTA-AM, indicating that was a Ca 2ϩ -dependent process. Furthermore, the Tf-stimulated exosome release was also inhibited by 2-APB, suggesting that IP 3sensitive Ca 2ϩ channels participate in this process. This is in agreement with a recent observation that the addition of apotransferrin to cultured oligodendroglial cells increased the levels of IP 3 (41). DISCUSSION In this study we have shown that the ionophore MON induces the formation of large MVBs and stimulates the release of the internal vesicles called exosomes. It has been shown that MON induces catecholamine secretion from adrenal chromaffin cells (14,42). The release of regulated secretory granules is known to be Ca 2ϩ -dependent. Here, we present evidence that the MON-stimulated exosome secretion in K562 cells is indeed a calcium-dependent event. The application of MON generated a marked elevation of Ca 2ϩ that was dependent on both an extracellular source and intracellular Ca 2ϩ stores. The role for Ca 2ϩ on exosome release was confirmed by the use of the Ca 2ϩ ionophore A23187. Similar results were obtained with agents known to alter the pH of vacuolar compartments such as chloroquine and bafilomycin, an inhibitor of the proton pump. Both compounds stimulated exosome release via a Ca 2ϩ -dependent mechanism, because the effect was abrogated by BAPTA-AM. We have evidence that chloroquine elevated intracellular Ca 2ϩ in a similar manner as MON. These data are in agreement with previous observations indicating that chloroquine causes a substantial Ca 2ϩ release in the parasite Plasmodium chabaudi (30). Similarly, it has been shown that bafilomycin alters cytosolic Ca 2ϩ by discharging intracellular Ca 2ϩ pools in lizard red blood cells (31). Therefore, taken together, our results clearly indicate that Ca 2ϩ is a key participant in the exosome release process. We believe that this is a relevant observation because, depending on their origin, exosomes can play roles in different physiological processes (8). For example, activated platelets release exosomes at sites of vascular injury where they may have a signaling/adhesion function (7). Antigen-presenting cells also secrete exosomes that carry peptide-loaded MHC molecules functioning as intercellular vehicles for antigenic material. Therefore, our observation that Ca 2ϩ regulates exosome release suggests that a signal transduction mechanism is likely involved in the activation of exosome-carrying cells to release these small vesicles at the proper site.
The use of MON allowed us to assess that Ca 2ϩ plays an important role in exosome secretion. It has been proposed that MON acts on acidic intracellular organelles such as endosomes and lysosomes, exchanging H ϩ for Na ϩ and causing swelling of these vesicles by passive water influx (44). Therefore, it would be possible that a similar mechanism might be involved in the generation of the large MVBs formed after MON treatment. However, our results indicate that the formation of the enlarged endosomes is a calcium-dependent event, because this process was completely abrogated by the Ca 2ϩ chelator BAPTA-AM. It is tempting to speculate that the formation of these gigantic endosomes is not only due to swelling of the vesicles but requires also the contribution of membranes from other sources, implying fusion among vesicular compartments. It is becoming evident that many intracellular transport events depend on Ca 2ϩ (45) and, thus, it is likely that Ca 2ϩ might be required for the fusion events involved in the generation of the enlarged MVBs. Experiments are underway to address this possibility.
MON, as a Na ϩ ionophore, transports a molecule of Na ϩ inside the cells per molecule of H ϩ transported to the extracellular media. The increase in intracellular Na ϩ activates the Na ϩ /Ca 2ϩ exchanger in a reverse mode, leading to an initial increase in cytosolic Ca 2ϩ (16). Our data are in agreement with a requirement for an early entrance of Na ϩ , because the MONmediated effects were completely abrogated by amiloride, an inhibitor of the H ϩ /Na ϩ and Na ϩ /Ca 2ϩ exchangers. Our data are also compatible with the idea that Na ϩ entry mediated by MON could activate the production of IP 3 , which in turn releases Ca 2ϩ from intracellular stores. Subsequently, the emptiness of these stores might trigger the opening of store-operated Ca 2ϩ channels (SOC) at the plasma membrane that induces a sustained Ca 2ϩ rise consistent with the requirement for extracellular Ca 2ϩ in our system. Data presented in this report indicate that IP 3 -dependent channels are involved in MON-stimulated exosome release, indicating that a transient Ca 2ϩ rise mediated by the stimulation of IP 3 receptors is critical for this event. However, the formation of the enlarged MVBs was only partially inhibited and not completely abrogated by inhibition of the IP 3 receptors, suggesting that an additional mechanism might be involved in the generation of these large endosomes.
Interestingly, we have observed that the large MVBs generated by MON-treatment are filled with Ca 2ϩ . It is likely that the MON-stimulated antiport activity might be exerted primarily at the plasma membrane, but it is also possible that a similar effect occurs on intracellular membranes. Indeed, it has been shown that MON selectively induces the secretion of azurophil granule contents by directly affecting the granule membrane (46). Swanson and co-workers (28) have reported that lysosomes contain high concentration of Ca 2ϩ and therefore could function as an intracellular Ca 2ϩ source. These authors have shown that changes in lysosomal pH resulted in the movement of vacuolar Ca 2ϩ out of lysosomes into the cytoplasm, possible via pH-dependent calcium channels or pumps. Our results indicate that, even though a thapsigargin-sensitive Ca 2ϩ pump is involved in the release of exosomes, inhibition of this pump does not block either the formation of the gigantic MVBs or the accumulation of Ca 2ϩ inside the large membranous structures. Therefore, it is likely that another type of Ca 2ϩ pump, present in intracellular compartments, might be responsible for filling the enlarged endosomes. PMR1, a Ca 2ϩ ATPase present in yeast Golgi (47) similar to the plasma membrane Ca 2ϩ ATPase (PMCA), is thapsigargin-insensitive. A Ca 2ϩ ATPase in the Golgi of mammary tissue (48) with characteristics slightly different from the PMCA and the endoplasmic reticulum Ca 2ϩ ATPase (SERCA) has also been identified. This mammary Golgi secretory pathway Ca ϩ2 -ATPase (SPCA) seems to be a homologue to the yeast PMR1 and is believed to be important for the function of the secretory pathway. Interestingly, we have recently shown that MVBs in K562 cells are formed, at least in part, by membrane influx from Golgi compartments (12). Furthermore, it has been shown that endocytic vesicles from reticulocytes possess a Ca 2ϩ ATPase that pumps Ca 2ϩ into the lumen of the vesicles (49). Therefore, it would not be unprecedented that a thapsigargin-insensitive Ca 2ϩ pump is involved in the formation of the enlarged MVBs and the accumulation of the lumenal Ca 2ϩ .
An interesting possibility is that the Ca 2ϩ present inside the MVBs is playing a critical role in the exosome secretory process. A role for intravesicular Ca 2ϩ in several secretory and intracellular fusion events has been proposed. Fusion of early endosomes seems to require the release of lumenal Ca 2ϩ for fusion to occur (50). Also, in insulin-secreting cells it has been shown that Ca 2ϩ depletion from granules inhibits exocytosis (21). Indeed, the presence of IP 3 receptors on insulin and somatostatin secretory granules has been demonstrated, suggesting that these organelles represent a readily mobilizable IP 3regulated Ca 2ϩ pool during the secretory process (52). As we mentioned above, it is known that MON disrupts proton gradients across the membranes, allowing the release of intravacuolar Ca 2ϩ . Therefore, it is possible that the localized release of Ca 2ϩ from the MVBs at the docking site may play an active role in the fusion complex/pore formation. Even though our results clearly indicate that Ca 2ϩ is a critical participant in the MON-stimulated exosome secretion, further experiments are required to elucidate whether the MVB luminal Ca 2ϩ participates in this process.
Finally, we have presented data indicating that Tf increased intracellular calcium in K562 cells and stimulated exosome release in a Ca 2ϩ -dependent manner, suggesting that the secretion of exosomes is under physiological control. Our results are consistent with previous observations that the binding of transferrin to its receptor raises intracellular Ca 2ϩ and stimulates receptor recycling in L2C cells (40). Also, transferrin recycling was stimulated by Ca 2ϩ in bovine chromaffin cells (53). Our results suggest that a component of the vesicular transport machinery involved in exosome secretion is regulated by Ca 2ϩ . Proteins such as synaptotagmins and calmodulin have been implicated in vesicular transport events where they function as Ca 2ϩ sensors (54). Therefore, Tf, by increasing intracellular Ca 2ϩ , may modulate some critical components of the transport/fusion machinery. Interestingly, in a recent paper it has been shown that the binding of apotransferrin to its receptor increased, via a calcium-calmodulin-dependent kinase, the levels of tubulin and actin, proteins known to be involved in vesicular trafficking (41). In conclusion, the TfR seems to function as a signal transduction molecule that modulates not only its own recycling but also its shedding via the exosome pathway, and Ca 2ϩ is one of the components of this signaling pathway triggered by Tf binding. It is important to mention that exosomes released to circulation from maturing red cells are the principal source of the soluble, circulating, truncated TfR (24). Soluble TfR levels 8-fold greater than normal have been reported in hemolytic anemias (51). Even though the physiological role of the soluble truncated TfR (43) has not been established yet, our observation that Tf stimulates exosome release and, as a consequence, the availability of soluble TfR, led us to speculate that this is a mechanism to regulate the free levels of circulating Tf.