Peroxisomes as Novel Players in Cell Calcium Homeostasis*

Ca2+ concentration in peroxisomal matrix ([Ca2+]perox) has been monitored dynamically in mammalian cells expressing variants of Ca2+-sensitive aequorin specifically targeted to peroxisomes. Upon stimulation with agonists that induce Ca2+ release from intracellular stores, peroxisomes transiently take up Ca2+ reaching peak values in the lumen as high as 50–100 μm, depending on cell types. Also in resting cells, peroxisomes sustain a Ca2+ gradient, [Ca2+]perox being ∼20-fold higher than [Ca2+] in the cytosol ([Ca2+]cyt). The properties of Ca2+ traffic across the peroxisomal membrane are different from those reported for other subcellular organelles. The sensitivity of peroxisomal Ca2+ uptake to agents dissipating H+ and Na+ gradients unravels the existence of a complex bioenergetic framework including V-ATPase, Ca2+/H+, and Ca2+/Na+ activities whose components are yet to be identified at a molecular level. The different [Ca2+]perox of resting and stimulated cells suggest that Ca2+ could play an important role in the regulation of peroxisomal metabolism.

Peroxisomes are quasi-ubiquitous organelles of eukaryotic cells that are involved in several metabolic pathways. They play an essential role in fatty acid ␣and ␤-oxidation, in the biosynthesis of ether phospholipids and bile acids, in the catabolism of purines and polyamines and in the degradation of hydrogen peroxides, prostaglandins, glyoxylate, and L-pipecolic acid (1). Impairment of peroxisomal activities causes "peroxisomal disorders," most of which are associated with severe neurological symptoms as in the Zellweger spectrum (1)(2)(3). Structurally, peroxisomes consist of a proteinaceous milieu limited by a single lipid bilayer originally thought to be freely permeable to small solutes. In contrast, recent reports demonstrated the existence of peroxisomal membrane transporters in both yeast and mammalian cells (4 -8), thus suggesting a strictly regulated activity of peroxisomal pathways acting in concert with cytosolic metabolism. In addition, the hypothesis of peroxisomes playing a direct role in intracellular signaling was supported (9), but no information until now is available on how extracellular agonists could have regulatory effects on peroxisomal biochemical pathways via their second messengers. In this context, we investigated for the first time the properties of peroxisomes in handling Ca 2ϩ , one of the most ubiquitous cellular second messengers.
In this work we provide direct evidence that peroxisomes play a role in Ca 2ϩ homeostasis by using the targeted recombinant aequorin approach that has been previously applied to other subcellular compartments such as mitochondria (10), nucleus (11), endoplasmic reticulum (ER) 4 (12), Golgi apparatus (13), and secretory vesicles (14). We generated two novel peroxisomally targeted aequorins, peroxAEQwt and perox-AEQmut, suitable for dynamic monitoring of Ca 2ϩ in intact cells over a wide range of concentrations. We found that a large transient Ca 2ϩ uptake occurs in peroxisomes of cells stimulated with extracellular agonists. Furthermore, in steady state conditions, Ca 2ϩ in peroxisomal lumen is maintained at concentrations ϳ20-fold higher than in cytosol. The sensitivity of peroxisomal Ca 2ϩ transport to a set of different ionophore reagents unravels the existence of an unexpectedly complex bioenergetic framework across the peroxisomal membrane, whereby a H ϩ gradient (in resting state) and H ϩ and Na ϩ gradients (in stimulated cells) sustain Ca 2ϩ uptake into the peroxisomal lumen.
Our work provides clear evidence that peroxisomes are involved in Ca 2ϩ homeostasis, thus adding further complexity to the intracellular network of Ca 2ϩ signaling. The dynamic flux of Ca 2ϩ ions across the peroxisomal membrane presented herein has unique characteristics when compared with previously investigated subcellular compartments, suggesting the existence of yet unidentified peroxisomal membrane transporting systems as well as the potential for Ca 2ϩ to play a role in the regulation of peroxisomal metabolism.

EXPERIMENTAL PROCEDURES
Construction of peroxAEQs and pHluorin cDNAs-Peroxisome-targeted wild-type and mutant aequorins (peroxAEQwt and peroxAEQmut) were generated by appending the PTS1 signal to their C termini. For the amplification of both aequorin variants, oligonucleotides 5Ј-CAT AAG CTT ATG TAT GAT GTT CCT GAT TAT-3Ј and 5Ј-TAA GAA TTC TTA TAA TTT GGA GGG GAC AGC TCC ACC GTA-3Ј were used as forward and reverse primers to insert the PTS1 sequence downstream the HindIII-EcoRI fragments encoding the HA1-tagged aequorin (15) and the HA1-tagged D119A aequorin mutant cDNAs (12). The final chimeric cDNAs were then shuttled as HindIII-EcoRI fragments into the mammalian vectors pcDNA3 (Invitrogen) (pcDNA3-peroxAEQwt and pcDNA3-peroxAEQmut; see Fig.  1A). The cDNAs encoding the cytosolic and peroxisomal variants of the ratiometric pH-sensitive pHluorin (16) were lifted from pHPR258 and pHPR282 plasmids, respectively (5), as EcoRI-SalI fragments and ligated into EcoRI-XhoI-cut pcDNA3 (pcDNA3-cyt-pHluorin and pcDNA3-peroxi-pHluorin). Escherichia coli DH5␣ strain was used for all plasmid amplifications and isolations. The sequences of the inserts were verified.
Cell Cultures and Transfection-All cell types (passages 15 and 30) used in the present work were grown in 75-cm 2 flasks in the presence of an appropriate medium (Sigma-Aldrich) supplemented with 10% fetal bovine serum. In particular, CHO cells were cultured in Ham's F-12 nutrient mixture; HeLa, HepG2, and COS-7 cells in Dulbecco's modified Eagle's medium; and HEK293 cells in Dulbecco's modified Eagle's medium/ F-12 medium. Twenty-four hours before transfection, the cells were seeded onto 13-mm (for aequorin measurements), 24-mm (for immunofluorescence) or 40-mm (for pHluorin measurements) glass coverslips and allowed to attain 40 -60% confluence. Transfection was carried out with 4, 8, or 12 g of total plasmid DNA for 13-, 24-, or 40-mm coverslips, respectively, according to a standard calcium-phosphate procedure (17), and the experiments were performed 36 -48 h after transfection.
Immunolocalization of the Peroxisomal HA1-tagged Recombinant Aequorins-For the immunofluorescence experiments, CHO cells were co-transfected with 4 g of pcDNA3-peroxAEQwt or pcDNA3-peroxAEQmut and 4 g of peroxi-DsRed2 plasmid (Clontech, Mountain View, CA) containing the coding sequence of the red fluorescent DsRed protein having a type-2 PTS signal (1) for its specific expression in the peroxisomes. Thirty-six to 48 h after transfection, the cells were fixed in the presence of formaldehyde and permeabilized with 0.1% Triton X-100, as previously described (13), and then incubated for 1 h at 37°C in a wet chamber with a 1:250 dilution (in phosphate-buffered saline) of the HA.11 monoclonal antibody, clone 16B12 (Covance, Princeton, NJ) raised against the HA1 epitope. Immunostaining of the cells was performed with the green fluorescent AlexaFluor 488 anti-mouse secondary antibody (Molecular Probes, Eugene, OR). Wide field fluorescences were acquired through a Zeiss Axiovert 200 microscope (Zeiss, Jena, Germany) equipped with a CoolSNAP HQ CCD camera (Roper Scientific, Trenton, NJ). The images were analyzed using the Metamorph software (Universal Imaging Corporation, Downington, PA) and processed with the AutoDeblur 2D deconvolution software (AutoQuant Imaging, Inc., Watervliet, NY).
Aequorin Reconstitution and Luminescence Measurements-Recombinant aequorins were reconstituted by incubating transfected cells with 5 M coelenterazine (for cytAEQwt (15), mitAEQmut (18), peroxAEQwt and peroxAEQmut), or 5 M coelenterazine n (for erAEQmut (12)) in KRB/Ca 2ϩ , pH 7.4, in 5% CO 2 atmosphere at 37°C for 2-3 h. When a reduction of the intracellular Ca 2ϩ content was necessary, aequorin reconstitutions were performed by incubating the cells for 1-2 h at 4°C in KRB in the absence of Ca 2ϩ and in the presence of 5 M coelenterazine, 5 M ionomycin, as a Ca 2ϩ ionophore, and 0.6 mM EGTA. After this treatment, the cells were extensively washed with KRB supplemented with 2% bovine serum albumin and 1 mM EGTA (13). Where indicated, the cells were permeabilized with 20 M digitonin for 1 min (19) and subsequently perfused with "intracellular buffer" supplemented with 0.1 mM EGTA (IB/EGTA: 140 mM KCl, 10 mM NaCl, 1 mM KH 2 PO 4 , 5.5 mM glucose, 2 mM MgSO 4 , 1 mM ATP, 2 mM succinate, 20 mM HEPES, pH 7.05) at 37°C. The Ca 2ϩ -dependent luminescence of aequorin was measured by perfusing the cells with KRB/ Ca 2ϩ supplemented with 5.5 mM glucose and stimulating them with 100 M ATP (CHO, HepG2, COS-7, and HEK293) or 100 M histamine (HeLa cells). In permeabilized cells, [Ca 2ϩ ] measurements were carried out perfusing the cells in IB without EGTA, i.e. ϳ5 M free Ca 2ϩ (13), and activating with 5 M IP 3 . All of the experiments were terminated by lysing the cells with 100 M digitonin in a hypotonic Ca 2ϩ -rich solution (10 mM CaCl 2 in H 2 O), thus discharging the remaining aequorin pool. The light signal was collected in a purpose-built luminometer and calibrated into [Ca 2ϩ ] values as previously described (20). When Ca 2ϩ sensitivity of peroxAEQs luminescence was studied, CHO cells expressing peroxAEQwt or peroxAEQmut were lysed by freezing/thawing cycles at Ϫ80°C, and the probes were reconstituted with 5 M coelenterazine. Lysed cells were then collected into a PerkinElmer MicroBeta Jet luminometer, and Ca 2ϩ sensitivity of the probes was measured as previously described (12) by relating the rate of the luminescence emitted in the presence of EGTA-buffered Ca 2ϩ solutions (L) to that emitted after complete consumption of the probes with excess of Ca 2ϩ (L max ).

Subcellular Targeting and Validation of Peroxisomal
Aequorins-Peroxisomal variants of the hemagglutinin (HA1)tagged wild-type (15) and D119A mutant (12) aequorins were generated by appending the PTS1 signal (1) to their C termini (Fig. 1A). The two different aequorin chimeras, designated as peroxAEQwt and peroxAEQmut, were expressed in mammalian cells, and their peroxisomal localization was confirmed by immunocytochemical experiments. The fluorescence of immunostained cells expressing peroxAEQwt (Fig. 1B, panel i) or peroxAEQmut (not shown) extensively overlapped with that of the co-expressed recombinant peroxisomal marker peroxi-DsRed2 (Fig. 1B, panel ii), whereas no significant overlay was imaged when cells co-expressed mitochondrial or ER fluorescent recombinant proteins (21) (not shown).
As previously reported (22), the addition of targeting motifs to the C terminus of aequorin could modify its stability as well as its Ca 2ϩ -dependent luminescence emission. Accordingly, the light emission from lysates of CHO cells expressing the peroxAEQs was measured upon reconstitution of the probes in the presence of various Ca 2ϩ -buffered solutions at neutral pH. The emitted luminescence of peroxisomal probes at free [Ca 2ϩ ] ranging between 10 Ϫ9 and 10 Ϫ2 M (L) was related to the total luminescence emitted in the presence of saturating [Ca 2ϩ ] (L max ), and the calculated L/L max values were actually similar to the output of lysates of cells expressing the cytosolic wild-type (cytAEQwt) or the mitochondrial mutant aequorin (mit-AEQmut) used in the same experimental conditions as already calibrated reference probes (11, 23) (data not shown). In addi-tion, because Ca 2ϩ sensitivity of aequorins depends on the pH of the surrounding medium, we determined the pH of the peroxisomal lumen using a peroxisomally targeted variant of the pH-sensitive protein pHluorin (16) (peroxi-pHluorin). When compared with a cytosolic pHluorin, the measured peroxisomal and cytosolic pH values showed minor differences that do not significantly affect aequorin Ca 2ϩ sensitivity (CHO cells, 6.92 Ϯ 0.15, n ϭ 25 in peroxisomes versus 7.08 Ϯ 0.11, n ϭ 25 in cytosol; HeLa cells, 7.09 Ϯ 0.25, n ϭ 15 in peroxisomes versus 7.23 Ϯ 0.09, n ϭ 15 in cytosol). In view of these data, the Ca 2ϩ response for peroxAEQs was therefore assumed to be equivalent to that of the cytosolic and mitochondrial variants.
Peroxisomes Transiently Accumulate Ca 2ϩ upon Agonist Stimulation-Evidence for a possible role of peroxisomes in cellular Ca 2ϩ signaling was obtained when peroxAEQmut was transiently expressed in mammalian cells, and its Ca 2ϩdependent luminescence was measured in the presence of agonists that elicit the IP 3 -mediated release of Ca 2ϩ from intracellular stores (24). Upon cell stimulation, a rapid transient Details of peroxisomal co-localization are magnified in the insets at the right corners of both panels. Bars, 10 m.

FIGURE 2. Ca 2؉ uptake occurs in peroxisomes of agonist-stimulated CHO cells to a higher extent than in mitochondria and cytosol and is not inhibited by ruthenium red.
A-C, cells were seeded onto glass coverslips and transfected in parallel batches with pcDNA3-peroxAEQmut, VR1012-mit-AEQmut (18), and VR1012-cytAEQwt (15). After 36 h, recombinant aequorins were reconstituted with 5 M coelenterazine in 5% CO 2 atmosphere at 37°C for 2 h in KRB/Ca 2ϩ medium supplemented with 5.5 mM glucose. At the luminometer, the cells were perfused with the same medium and triggered with 100 M ATP. The experiments were terminated lysing the cells with a Ca 2ϩ -rich hypotonic medium, and collection and calibration of the output luminescence were carried out, as described under "Experimental Procedures. " The data are representative of 15 experiments that yielded similar results. D, CHO cells were transfected in parallel with VR1012/ mitAEQmut or pcDNA3/peroxAEQmut. After reconstitution of the recombinant probes with 5 M coelenterazine, the cells were permeabilized by 1-min incubation with 20 M digitonin (19) and perfused in IB buffer. Mitochondrial (mit) and peroxisomal (perox) Ca 2ϩ uptakes were measured upon the addition of 5 M IP 3 in the presence or absence of 4 M ruthenium red. Aequorin luminescence was calibrated and measured into [Ca 2ϩ ] values, and the presented data Ϯ S.E. (error bars) from four replicates express the inhibition percentages of Ca 2ϩ uptake as compared with control cells (*, p Ͻ 0.001, oneway analysis of variance followed by Bonferroni's t test).
Ca 2ϩ influx occurred in peroxisomes ( Fig. 2A), parallel to the [Ca 2ϩ ] increases recorded in mitochondria and cytosol (Fig. 2, B and C). The kinetics of the peroxisomal [Ca 2ϩ ] changes were similar to those already described for mitochondria (23). How-ever, as shown in Table 1 for several stimulated cell lines, the measured agonist-induced peroxisomal [Ca 2ϩ ] peak values ([Ca 2ϩ ] perox ) varied from ϳ50 to ϳ100 M, and they were significantly and reproducibly higher than those measured in mitochondria and were unaffected by ruthenium red (Fig. 2D), a well known inhibitor of mitochondrial Ca 2ϩ uptake (23). Therefore, peroxisomes transiently reach a higher [Ca 2ϩ ] than mitochondria, peaking as rapidly as mitochondria although through a different molecular route.
Peroxisomal Ca 2ϩ Uptake in Stimulated CHO Cells-To investigate the mechanism leading to the large agonist-evoked [Ca 2ϩ ] perox increase, we tested the effect of reagents dissipating pH gradients across biological membranes (Fig. 3). In ATPchallenged CHO cells expressing peroxAEQmut, the clear reduction of [Ca 2ϩ ] perox peaks induced by chloroquine or the mitochondrial uncoupler FCCP (untreated, 98 Ϯ 10 M, n ϭ 15; chloroquine-treated, 78 Ϯ 6 M, n ϭ 10; FCCP-treated, 58 Ϯ 5 M, n ϭ 10; Fig. 3A (Fig. 3A). The rapid peroxisomal Ca 2ϩ uptake in ATP-stimulated CHO cells could therefore be led by gradients of both H ϩ and Na ϩ ions. It should be noted that FCCP, but not monensin, reduced mitochondrial Ca 2ϩ uptake (Fig. 3B); however, its inhibitory effect was significantly more pronounced in mitochondria than

TABLE 1
Agonist-induced ͓Ca 2؉ ͔ peak values measured in cytosol, mitochondria, and peroxisomes of different cell lines All of the indicated cells were transfected in parallel with VR1012/cytAEQwt, VR1012/mitAEQmut, and pcDNA3/peroxAEQmut plasmids, which respectively encode cytosolic, mitochondrial, and peroxisomal aequorin chimeras. The recombinant probes were reconstituted after 36 h with 5 M coelenterazine in 5% CO 2 atmosphere at 37°C for 2 h in KRB/Ca 2ϩ medium plus 5.5 mM glucose. At the luminometer, the cells were perfused in the same medium and stimulated with 100 M ATP, except the HeLa cells, which were triggered with 100 M histamine. The data represent the maximum of ͓Ca 2ϩ ͔ peaks measured upon agonist stimulation. ] perox changes, we monitored the peroxisomal pH in ATP-stimulated CHO cells expressing the ratiometric pH-sensitive probe pHluorin. As shown in Fig. 4, upon cell stimulation peroxisomal Ca 2ϩ influx occurs contemporary with a bland alkalinization of peroxisomal pH soon followed by a more marked and protracted transient acidification mainly associated to the subsequent Ca 2ϩ efflux. According to these data, if the Ca 2ϩ influx may take part along with a hardly monitorable H ϩ efflux, the subsequent peroxisomal acidification of the lumen suggests that Ca 2ϩ re-extrusion occurs, directly or indirectly, in exchange with H ϩ . Such a prolonged compensatory response of the organelle in part recalls what happens in mitochondria where transitory alkalinization of the mitochondrial matrix occurs when Ca 2ϩ -mobilizing agents are added (29).

Cells
Peroxisomal Ca 2ϩ Content in Resting Conditions-The experiments performed using peroxAEQmut revealed a low number of total luminescence counts as compared with those recorded with cytAEQwt or mitAEQmut (Fig. 5). The possibility that peroxisomes concentrate Ca 2ϩ in their lumen during resting conditions, consuming the peroxisomally reconstituted aequorin (13), was tested in CHO cells expressing the probe peroxAEQwt, endowed with higher Ca 2ϩ affinity (30) and reconstituted after depleting the intracellular organelles of calcium (13) (Fig. 6). At the luminometer, when Ca 2ϩ -depleted cells were refilled with Ca 2ϩ , after a lag of ϳ30 s, [Ca 2ϩ ] perox increased up to ϳ3 M, and then the signal decreased and after 1 min virtually leveled off for several minutes (Fig. 6A). By contrast, when the same experiment was performed using peroxAEQmut, the [Ca 2ϩ ] changes were not detectable (Fig.  6B) because of the lower sensitivity of the probe. Therefore, in the absence of Ca 2ϩ -releasing agonists, peroxisomes are able to accumulate [Ca 2ϩ ] 20 -30-fold higher than in cytosol (15). By completing the experiments, the addition of ATP after Ca 2ϩ refilling induced [Ca 2ϩ ] perox peaks entirely similar to those obtained in cells that had not been treated with EGTA and ionomycin, when expressing both peroxAEQmut and peroxAEQwt (Fig. 6, B and C), demonstrating the functional integrity of the organelles after Ca 2ϩ depletion treatment.
Pathways of Peroxisomal Ca 2ϩ Accumulation in Resting Conditions-Ca 2ϩ accumulation in peroxisomes in resting was first investigated using thapsigargin and tBuBHQ (31,32), specific inhibitors of sarco/endoplasmic Ca 2ϩ -ATPase that actively pump Ca 2ϩ from the cytosol to the ER lumen (33). When Ca 2ϩ was added back to Ca 2ϩ -depleted cells that had been preincubated with both thapsigargin (Fig. 7A) or tBuBHQ (not shown), Ca 2ϩ uptake into the ER was strongly reduced. By contrast, under the same experimental conditions, Ca 2ϩ entry into peroxisomes was markedly increased (Fig. 7B), a likely finding because of the enhanced [Ca 2ϩ ] cyt recorded in parallel (not shown) and already previously reported (13). Similar results were also obtained when digitonin-permeabilized Ca 2ϩdepleted cells were Ca 2ϩ refilled in the presence of orthovanadate, a wide spectrum inhibitor of P-type Ca 2ϩ -ATPase (Fig.  7C), thus excluding a role of Ca 2ϩ -ATPase in resting peroxisomal Ca 2ϩ homeostasis. Furthermore, the addition of ATP to thapsigargin-treated cells (Fig. 7B) or IP 3 to orthovanadate-treated cells (Fig. 7C) did not induce any further peroxisomal Ca 2ϩ uptake in contrast with what was observed in nontreated cells. This result is clearly due to ER Ca 2ϩ depletion and confirms the relationship between peroxisomes and the agonist-evoked Ca 2ϩ release from intracellular stores.
Interestingly, when Ca 2ϩ -depleted CHO cells were incubated with bafilomycin concentrations inhibiting the V-type ATPases, but not the P-type ATPases (34), a clear decrease of peroxisomal Ca 2ϩ refilling was observed (Fig. 7D). This finding further supports the suggestion that a putative peroxisomal V-type ATPase activity occurs in CHO cells. Consequently, we tested the effect of FCCP with the result that peroxisomal Ca 2ϩ refilling was virtually abolished (Fig. 7E). Moreover, FCCP caused a release of Ca 2ϩ from peroxisomes that had been reloaded with Ca 2ϩ (Fig. 7F). Therefore, during resting a H ϩ gradient is not only required for peroxisomal Ca 2ϩ uptake but also to maintain the existing Ca 2ϩ concentration gradient between peroxisomes and cytosol.
Ca 2ϩ Sensitivity of Peroxisomes-The two different kinetics of peroxisomal Ca 2ϩ uptake respectively shown in resting and in agonist-stimulated CHO cells may unravel a different sensitivity of these organelles when [Ca 2ϩ ] cyt changes. The response of peroxisomes to different buffered [Ca 2ϩ ] was firstly studied in permeabilized CHO cells. As presented in Fig. 8 (A and B), the perfusion with 5 M and 10 M Ca 2ϩ induced an increase of [Ca 2ϩ ] perox according to a kinetic comparable with that monitored during Ca 2ϩ refilling in intact cell (Fig. 6)  These results suggest that a basal peroxisomal Ca 2ϩ transport is activated in the presence of low [Ca 2ϩ ] cyt (resting cells), whereas a different transport system is triggered when high [Ca 2ϩ ] cyt occurs (stimulated cells). Indeed, when intact CHO cells after Ca 2ϩ refilling were treated with tBuBHQ (Fig. 8, C  and D) that promotes a slow release of Ca 2ϩ from the ER and a small increase of the [Ca 2ϩ ] cyt , it did not produce any increase in mitochondrial Ca 2ϩ (not shown) as already described (23), whereas in peroxisomes it only minimally enhanced the Ca 2ϩ content, thus confirming that the agonist-induced peroxisomal Ca 2ϩ uptake is mainly related to the higher InsP 3 -mediated [Ca 2ϩ ] cyt microdomains.

DISCUSSION
The relatively new concept of a selective metabolic communication between cytosol and peroxisomes raises questions concerning whether and how peroxisomes are involved in cellular signaling pathways. In the present work, we focused on the possibility that peroxisomes take part in the intracellular homeostasis of Ca 2ϩ , a versatile second messenger that is able to decode a great variety of extracellular stimuli into the simultaneous control of different biochemical pathways occurring in various subcellular compartments. To the best of our knowl-edge, no information about the involvement of peroxisomes in Ca 2ϩ signaling has hitherto been reported.
With the aim of investigating the Ca 2ϩ concentration in peroxisomes, we efficiently targeted to the peroxisomal lumen recombinant aequorins. Our recordings provide the evidence that peroxisomal Ca 2ϩ handling is substantially different from those already reported for other subcellular organelles (33). In particular, mammalian peroxisomes stably maintain Ca 2ϩ in their lumen during resting up to concentrations ϳ20-fold higher than in cytosol (ϳ2 versus ϳ0.1 M), whereas with challenging the cells with Ca 2ϩ -releasing agonists, a more conspicuous although transient peroxisomal Ca 2ϩ uptake (up to ϳ100 M) is induced. Also, the agonist-induced [Ca 2ϩ ] perox peaks are characterized by a rapid entry and a subsequent slower efflux of the ion, compatible in shape but not in amplitude to those monitored in mitochondria (10) with a different sensitivity to pharmacological treatments. According to these data, peroxisomes, which represent only a very low percentage of the cell volume (35), are unlikely candidates to be effective stores of intracellular calcium (36), and the relatively low intraperoxisomal Ca 2ϩ concentration during resting could reflect the presence of a driving force across the peroxisomal membrane. This hypothesis was pointed out when a complete suppression and ϳ50% inhibition of peroxisomal Ca 2ϩ uptakes during resting and agonist stimulation, respectively, were produced in the presence of FCCP that dissipates pH gradients across biological membranes. This suggested the existence of a H ϩ gradient across the peroxisomal membrane and, in bioenergetic terms, made peroxisomes similar to mitochondria where Ca 2ϩ movements across their membrane depend strictly on the large H ϩ electrochemical gradient and FCCP is one of the best described inhibitors (23). More specifically, in peroxisomes FCCP generated an immediate inhibitory effect on the slower and more contained Ca 2ϩ uptake during resting. By contrast, upon cell stimulation when a much larger and faster Ca 2ϩ accumulation takes place, the partial FCCP effect could only be monitored after a short (ϳ10 s) preincubation, 5 which is different from what happens in mitochondria where an almost complete inhibition of Ca 2ϩ uptake occurs. These observations denoted that in resting but not during cell stimulation, a H ϩ gradient is by itself sufficient to sustain the peroxisomal Ca 2ϩ uptake, thus representing the only determinant for the resting Ca 2ϩ accumulation.
On the basis of the differences of [Ca 2ϩ ] recorded between cytosol and peroxisomal lumen, the driving force required for the peroxisomal Ca 2ϩ entry would be of ϳ30 and ϳ50 mV in resting and stimulated cells, respectively, according to the Nernst equation. Actually, in resting cells, by monitoring pHluorin ratiometric fluorescence, we measured a non-significant ⌬pH between cytosol and peroxisomes of ϳ0.2, corresponding to a ⌬ H of ϳ12-15 mV that is apparently insufficient to sustain peroxisomal Ca 2ϩ accumulation. In addition, the agonist-evoked Ca 2ϩ influx into peroxisomes is accompanied only by a modest increase of peroxisomal pH, which would exclude a possible Ca 2ϩ /H ϩ exchange across the peroxisomal membrane. In reality, it should be noted that in our experimen- tal conditions, the pH sensitivity of the probe pHluorin does not allow accurate determination of ⌬pH values less than ϳ0.3-0.4. Furthermore, it might be also inferred that a clear difference between cytosolic and peroxisomal pH might be only displayed in microdomains localized juxtaposed to the peroxisomal membrane (37) and that a Na ϩ gradient could be part of the driving force for Ca 2ϩ accumulation (see below). Finally, pH buffering properties of the peroxisomal matrix are suggested by the reversible acidification of peroxisomal lumen following agonist stimulation (Fig. 4), in which repolarization mechanisms could be related to a different orientation of H ϩ ions across the peroxisomal membrane, if compared with mitochondria (29).
For the completion of the larger agonist-evoked peroxisomal Ca 2ϩ increases, the main energetic contribution is most likely represented by a Na ϩ gradient. When the Na ϩ /H ϩ ionophore monensin is applied to stimulated cells, the inhibitory effect on peroxisomal Ca 2ϩ uptake is higher than with FCCP, and the Ca 2ϩ peaks are almost completely abolished with the contemporary presence of both reagents. These data suggest that the total driving force for Ca 2ϩ entry into peroxisomes derives from the sum of a H ϩ gradient, ⌬ H, and, to a higher extent, a Na ϩ gradient, ⌬ Na. The peroxisomal H ϩ gradient might be generated by a peroxisomal V-type H ϩ -ATPase. In fact, the V-type H ϩ -ATPase inhibitor bafilomycin (25) clearly inhibits peroxisomal Ca 2ϩ uptake, both in resting and in activated cells.
Although the presence of this complex in peroxisomal membrane has already been hypothesized by others (38 -40), further investigation is required to validate this hypothesis. Alternatively, the observed partial inhibition of peroxisomal Ca 2ϩ uptake by bafilomycin might be secondary to the inhibition exerted by bafilomycin on the V-ATPase of other subcellular compartments, particularly endosomes, Golgi apparatus, and vacuoles (26), because release of H ϩ from these compartments subsequent to bafilomycin treatment (41-43) would oppose a peroxisomal Ca 2ϩ /H ϩ exchange by dissipating the H ϩ gradient across the peroxisomal membrane.
Although the existence of ⌬pH across the peroxisomal membrane still represents an unsolved question (5, 39, 44 -46) and the data reported herein on the presence of a peroxisomal Na ϩ gradient deserve further confirmation, in view of our results, we assume that H ϩ and Na ϩ ions could be more distributed on the internal side of the peroxisomal membrane, thus allowing the uptake of Ca 2ϩ into peroxisomes in exchange with H ϩ (in resting) and also with Na ϩ (upon cell stimulation). Accordingly, we hypothesize the existence in the peroxisomal membrane of Ca 2ϩ /H ϩ and Ca 2ϩ / Na ϩ exchangers having different Ca 2ϩ affinities, which would transport Ca 2ϩ against a concentration gradient into the peroxisomal lumen at the expense of the H ϩ and Na ϩ gradients, respectively. The high affinity Ca 2ϩ /H ϩ exchanger would mainly catalyze Ca 2ϩ uptake in the presence of low Ca 2ϩ concentrations in the cytosol in resting conditions, whereas the low affinity Ca 2ϩ /Na ϩ exchanger would catalyze the high peroxisomal Ca 2ϩ peaks in the presence of higher cytosolic Ca 2ϩ concentrations following Ca 2ϩ release from intracellular stores upon cell stimulation. In support of our assumption, if in stimulated cells CGP37157, an inhibitor of the Ca 2ϩ /Na ϩ exchangers (28,47), enhances Ca 2ϩ uptake in mitochondria, in peroxisomes it inhibits Ca 2ϩ uptake, thus clearly indicating the existence of a peroxisomal Ca 2ϩ /Na ϩ exchanger. Therefore, on the basis of an opposite orientation of ion gradients across the respective membranes, in peroxisomes Ca 2ϩ /H ϩ and Ca 2ϩ /Na ϩ exchangers would catalyze the entry of Ca 2ϩ toward the lumen, whereas mitochondrial Ca 2ϩ /H ϩ and Ca 2ϩ /Na ϩ function to discharge Ca 2ϩ from the mitochondrial matrix (23).
As to the decay of the agonist-evoked [Ca 2ϩ ] perox rise, we cannot determine whether it depends on a Ca 2ϩ buffering phenomenon within the peroxisomal lumen or on extrusion of Ca 2ϩ mediated by an unknown transport system. However, it seems evident that peroxisomal Ca 2ϩ homeostasis is accomplished through specific pathways requiring membrane transporters not yet identified at a molecular level. Consequently, the recent notion of a selective peroxisomal membrane with distinctive bioenergetics appears strengthened. Indeed, the [Ca 2ϩ ] perox values we have measured exclude the possibility of a peroxisomal membrane being freely permeable to ions. The large amplitude of the [Ca 2ϩ ] perox increase occurring when the cells are stimulated or when peroxisomes are perfused with high [Ca 2ϩ ] could suggest that peroxisomes are exposed to microdomains generated upon stimulation in proximity of ER Ca 2ϩ channels, similar to what was reported for mitochondria. Indeed, close contacts between peroxisomes and subcellular Ca 2ϩ stores have been documented (48), leaving open the possibility that the increase in [Ca 2ϩ ] perox could be heterogeneous and a subset of closely interacting peroxisomes could significantly increase the average [Ca 2ϩ ] perox peaks. Knowledge of the molecular entities responsible for the Ca 2ϩ transport across the peroxisomal membrane would elucidate whether the amplitudes of peroxisomal Ca 2ϩ accumulations both in resting and in activated cells could be related not only to different transporters with different affinities for Ca 2ϩ , but also whether their transport properties imply uniport mechanisms, likewise in mitochondria (33), or Ca 2ϩ exchange with counter ions, as our data strongly suggest. To date, as for mitochondrial Ca 2ϩ transport systems (33), scarce information is available about the molecular identity of putative peroxisomal Ca 2ϩ transporters. A secretory pathway Ca 2ϩ /Mn 2ϩ -ATPases isoform has been reported in peroxisomes (49), but our data with orthovanadate and thapsigargin rule out a role for any Ca 2ϩ -ATPase in Ca 2ϩ accumulation in peroxisomes. Alternatively, the function of Ca 2ϩ transporters or ion exchangers could be directly studied in vitro with peroxisomal membranes, but the difficult isolation of these organelles from cells because of their extreme fragility and low abundance could make these studies experimentally limiting.
In conclusion, we have shown that peroxisomes also take part in cellular Ca 2ϩ homeostasis. In analyzed cells, the intervention of several different pathways is necessary for peroxisomal Ca 2ϩ handling, thus reflecting complex bioenergetics that characterize peroxisomal membrane selectivity. Having excluded a role for peroxisomes as a subcellular Ca 2ϩ reservoir, future studies are warranted to biochemically characterize transporters and targets of peroxisomal Ca 2ϩ and to ascertain a regulatory role of the ion in peroxisomal function and its involvement in peroxisome-related disorders.