Constitutive calcium-independent release of Toxoplasma gondii dense granules occurs through the NSF/SNAP/SNARE/Rab machinery.

The signals and the molecular machinery mediating release of dense matrix granules from pathogenic protozoan parasites are unknown. We compared the secretion of the endogenous dense granule marker GRA3 in Toxoplasma gondii with the release of a stably transfected foreign reporter, beta-lactamase, that localizes to parasite dense granules. Both proteins were released constitutively in a calcium-independent fashion, as shown using both intact and streptolysin O-permeabilized parasites. N-Ethylmaleimide and recombinant bovine Rab-guanine dissociation inhibitor inhibited beta-lactamase secretion in permeabilized parasites, whereas recombinant hamster N-ethylmaleimide-sensitive fusion protein and bovine alpha-SNAP augmented release. Guanosine 5'-3-O-(thio)triphosphate, but not cAMP, augmented secretion in the presence but not in the absence of ATP. The T. gondii NSF/SNAP/SNARE/Rab machinery participates in dense granule release using parasite protein components that can interact functionally with their mammalian homologues.

The signals and the molecular machinery mediating release of dense matrix granules from pathogenic protozoan parasites are unknown. We compared the secretion of the endogenous dense granule marker GRA3 in Toxoplasma gondii with the release of a stably transfected foreign reporter, ␤-lactamase, that localizes to parasite dense granules. Both proteins were released constitutively in a calcium-independent fashion, as shown using both intact and streptolysin O-permeabilized parasites. N-Ethylmaleimide and recombinant bovine Rab-guanine dissociation inhibitor inhibited ␤-lactamase secretion in permeabilized parasites, whereas recombinant hamster N-ethylmaleimide-sensitive fusion protein and bovine ␣-SNAP augmented release. Guanosine 5-3-O-(thio)triphosphate, but not cAMP, augmented secretion in the presence but not in the absence of ATP. The T. gondii NSF/SNAP/SNARE/Rab machinery participates in dense granule release using parasite protein components that can interact functionally with their mammalian homologues.
Toxoplasma gondii is an obligate intracellular parasite that survives in all nucleated cells by creating a unique membrane bounded compartment called the parasitophorous vacuole (reviewed in Ref. 1). Like all Apicomplexan parasites, T. gondii contains three morphologically distinct secretory organelles: rhoptries, micronemes, and dense granules (DG). 1 The protein content of these organelles, the temporal regulation of their secretion, and the eventual localization within the infected cell of proteins secreted from the organelles are all different. T. gondii, therefore, represents an interesting model for analysis of the differential control of secretion from morphologically distinct secretory compartments. DG in T. gondii contain a homogeneous dense core and are similar in morphology to dense matrix granules in exocrine and neuroendocrine cells (reviewed in Ref. 2). DG exocytosis involves a putative fusion event with either the parasite plasma membrane or the inner membrane complex (a patchwork of closely opposed flattened vesicles located beneath the plasma membrane). In contrast to micronemes and rhoptries, which discharge their contents only at the time of cell invasion (3)(4)(5), the kinetics of DG exocytosis are less well defined. The tacit assumption (based largely on their morphology) has been that DG release is also a regulated process. We and others have described a burst of DG exocytosis shortly after invasion (3,6), but exocytosis of DG proteins also continues during the prolonged intracellular residence of the parasite (6 -8). Spontaneous release of DG proteins from extracellular parasites is also observed (9), and this release may be augmented by the addition of heat-inactivated serum (10). Thus, based on the limited information available, DG exocytosis exhibits characteristics of both a regulated and a constitutive process.
Phylogenetically, components of the NSF/SNAP/SNARE/Rab machinery are broadly conserved across species barriers in higher eukaryotes (27). To date, however, the identification of these molecules in pathogenic protozoa has been largely limited to monomeric GTPases of the Rab family. Functional analyses of their participation in granule exocytosis has not been reported. Experiments in the nonpathogenic ciliate, Paramecium (28,29), have identified Rab proteins on secretory vesicles and a SNARE-like molecule that complements a secretion mutant (30).
We recently reported that foreign soluble reporter proteins (Escherichia coli ␤-lactamase (BLA) and alkaline phosphatase) stably expressed in T. gondii are routed quantitatively to DG (31), suggesting that DG are a constitutive rather then regulated secretory organelle. Our observation raised several key questions. Is DG release constitutive or regulated? If DG release is regulated, what are the triggers for release? What is the protein machinery controlling DG docking and fusion? Is the protein secretory machinery conserved between T. gondii and higher eukaryotes?
We have addressed these questions by following the release of both an endogenous DG marker (GRA3) and the secreted recombinant reporter protein BLA in intact and streptolysin O-permeabilized extracellular parasites. Our results demonstrate that DG markers are released constitutively in a calcium-independent fashion, via the participation of the NSF/ SNAP/SNARE/Rab machinery. While calcium-independent constitutive release of soluble proteins from a dense matrix granule is unusual, the protein machinery mediating release in T. gondii is broadly conserved with that in higher eukaryotic cells.

Parasite and Cell Culture
The RH strain of T. gondii and a stable transgenic clone of the RH strain, expressing the soluble foreign secretion reporter BLA were described earlier (31). A stable transgenic clone of the RH strain expressing a cytosolic form of green fluorescent protein (GFP) (34) was used to assess cytosol leakage from the parasites permeabilized with SLO. A clone of the RH strain stably expressing the cytosolic marker E. coli ␤-galactosidase (35) was obtained from J. Boothroyd (Stanford, CA).
Parasites were maintained in monolayers of either Vero or HFF cells. Extracellular parasites were harvested, washed, and counted as described previously (9). In some experiments, extracellular parasites were incubated in calcium-and magnesium-free PBS (dissociation buffer) on ice for 30 min to remove network material adhered to parasites (36). After this incubation, parasites were washed twice with ice-cold AISS buffer and resuspended in the same buffer for secretion assays. In several experiments, HBSS buffer was used in place of AISS.

Secretion Assays
Secretion by extracellular parasites was determined by incubating 0.4 -1 ϫ 10 8 parasites/ml (total of 3 ml) in various buffers at 37°C. Glucose was essential for maximum secretion (secretion was reduced by 50% less in buffer without glucose; data not shown).
All assays were performed in prechilled Eppendorf tubes previously coated with BSA (1 mg/ml) to prevent parasites adhesion. Before transferring parasite suspensions from ice to 37°C for the secretion assay, 500 l of parasite suspension was removed and processed as the time 0 reading. The rest of the parasite suspension was incubated at 37°C in a water bath. After 15, 30, 60, and 120 min of incubation, 500 l of the parasite suspension was removed and centrifuged at 760 ϫ g for 10 min at 4°C to pellet intact parasites. The pellet was set aside, and the supernatant was further centrifuged at 7000 ϫ g for 10 min at 4°C. GRA3 in both the 7000 ϫ g supernatant and the pellet were determined by antigen capture sandwich ELISA (see below) and by immunoblot, as described previously (37). BLA activity was measured as described below.
Quantitative ELISA for GRA3 and ROP2,3 An ELISA was developed to measure GRA3 and ROP2,3 released from extracellular parasites in secretion assays. Plates (Linbro/Titertek) were coated with 50 l of monoclonal anti-GRA3 (T62H11) diluted 1:1000 in 0.05 M carbonate-bicarbonate buffer, pH 9.6, and incubated overnight at 4°C. The plates were washed four times with PBS/Tween 20/BSA. In initial experiments, whole parasite lysates were tested to validate the assay. A dose-response relationship was observed between the lysate from the increasing number of parasites and OD for GRA3 (data not shown).
To measure GRA3 released by intact parasites, supernatant derived from 10 8 parasites/ml was diluted 1:5 with PBS/Tween 20/BSA, and 100 l (2 ϫ 10 6 parasite equivalents) was added to each well and incubated at room temperature for 3 h. The plates were washed six times, and 100 l of rabbit anti-GRA3 or a control antiserum (1:2000) was added to each well and incubated at room temperature for 2 h. Plates were washed, incubated for an additional 2 h with 100 l of rabbit anti-goat IgG (Fc)-alkaline phosphatase (Promega) (1:4000 in PBS/Tween 20/ BSA), washed again, and developed with 200 l of p-nitrophenyl phosphate for 45 min followed by the addition of 50 l of 3 M NaOH. Plates were read at 550 nm in a Titertek Multiskan ELISA reader, and OD values for each sample were calculated by subtracting the average of triplicate wells with rabbit anti-GRA3 from the average of triplicate wells that received normal rabbit serum.
Total GRA3 and ROP2,3 in the parasite pellet were determined by lysing parasites (10 8 /ml) in a French press cell at 10,000 p.s.i. The parasite lysate was centrifuged at 12,500 ϫ g for 10 min. Serial 10-fold dilutions of the 12,500 ϫ g supernatant were prepared in PBS/Tween 20/BSA, and cellular GRA3 and ROP2,3 were quantitated as above.
Standard curves for GRA3 and ROP2,3 were generated by plotting net OD versus parasite number. The percentage release of GRA3 and ROP2,3 in the supernatant during secretion were calculated from the standard curve.

Enzyme Assays for BLA and ␤-Galactosidase
BLA in the parasite supernatant as well as in the pellet was determined by the method described earlier (38) with minor modifications. Parasite supernatant (4 ϫ 10 7 parasites/ml) diluted 2-fold with AISS buffer and 20 l was added to each well of a 96-well plate. The reaction was developed by adding 180 l of nitrocefin mix (0.2 mM nitrocefin, 0.25 mg/ml BSA, 50 mM potassium phosphate, 0.5% Me 2 SO, pH 7.0) for 30 min at 25°C followed by the addition of 50 l of 3 M NaOH. Plates were read at 492 nm in an ELISA reader, and the OD values for each sample were calculated by subtracting the average of duplicate wells with supernatant derived from RH parasites and wells with substrate only.
In preliminary experiments, a linear relationship was observed between BLA release and parasite number (data not shown). The enzymatic activity of released BLA was not affected by treatment with dithiothreitol, indicating that the oxidation-reduction state of the enzyme did not influence BLA activity (39). BLA activity in the pellet was substantially lower than in the supernatant, regardless of whether parasites were lysed in TX-100 or in the French press cell. Mixing experiments of pellet and supernatant indicated that there was an inhibitor of BLA activity in the parasite pellet (data not shown). Therefore, total released BLA was determined by immunoblot.
␤-Galactosidase activity in the supernatant as well as in the parasite pellet from ␤-galactosidase-expressing parasites was determined as described previously (35).

Influence of Incubation Conditions on DG Secretion in Extracellular Parasites
Temperature, Energy, and Serum Dependence of Secretion-Secretion was measured during incubation of extracellular parasites at 4 and 15°C, or with 10 mM sodium azide (NaN 3 ) at 15°C for 15 min, followed by incubation at 37°C for 1 h. The effect of serum on DG secretion by extracellular parasites was determined by adding 10% heat-inactivated fetal bovine serum to AISS buffer. The details of other conditions tested are described in the legend to Table I.

Parasite Permeabilization with SLO and Cytosol Depletion
SLO-Permeabilization of extracellular parasites with SLO was done using modifications of a described protocol (40). Extracellular parasites were washed two times with ice cold potassium acetate buffer containing glucose (115 mM potassium acetate, 2.5 mM MgCl 2 , 10 mM glucose, and 25 mM HEPES, pH 7.2, with KOH) and then incubated with 100 l of 1 unit/ml of SLO for 10 min at 4°C. The suspension was transferred to 37°C and incubated for 2 min, and then parasites were pelleted at 760 ϫ g for 5 min. Supernatant was removed and saved to measure released BLA during the permeabilization step. The parasite pellet was gently suspended in ice cold potassium acetate buffer, washed two times, and resuspended to 4 ϫ 10 7 parasites/ml. Permeabilized parasites were immediately used for secretion assays at 37°C. BLA secretion was also monitored from control parasites, which were treated similarly at every step without the presence of SLO.
Monitoring Permeabilization-Parasite permeabilization was assessed by incubating organisms with 4 g/ml propidium iodide for 5 min at room temperature, and the percentage of parasites positive for nuclear staining was quantitated by fluorescence microscopy.
Determining Cytosol Leakage-Release of lactate dehydrogenase as a cytosolic marker from parasites could not be quantitated with accuracy, due to contamination with host cell lactate dehydrogenase. Therefore, a stable transgenic clone of parasites expressing a cytosolic GFP marker (34) was treated with SLO. The percentage loss of GFP fluorescence, immediately after permeabilization and following continued incubation at 15°C for 1 h, was determined by measuring the fluorescence of parasites at 490 nm and emission of 510 nm in a fluorimeter and compared with total GFP in nonpermeabilized parasites treated in a similar fashion in the absence of SLO. The GFP release from SLOpermeabilized parasites was also assessed by phase and by fluorescence microscopy.

Effects of Calcium on DG Release from Permeabilized Parasites
Extracellular parasites were washed two times with cold potassium acetate-EGTA (5 mM) buffer and permeabilized with SLO in the same buffer, and then various concentrations of free calcium (0 -100 M final concentration) were added in the presence or absence of an ATP-regenerating system (ARS; 2 mM NaATP, 20 mM creatine phosphate, and 40 units/ml creatine phosphokinase). The concentrations of free Mg 2ϩ and Ca 2ϩ in the buffer were calculated using the software program of Foehr and Warchol (Ulm, Germany). Alternatively, parasites were permeabilized with SLO in potassium acetate buffer containing no EGTA and then preincubated with or without MAPTA-AM (50 M final concentration) in the presence or the absence of ARS at 0°C for 15 min followed by incubation at 37°C for 30 min for secretion to occur.

Determining the Protein Machinery of DG Secretion Using SLOpermeabilized T. gondii
The influence of N-ethylmaleimide on DG secretion was carried out by preincubating both SLO-permeabilized as well as nonpermeabilized parasites with or without 1 mM N-ethylmaleimide (NEM) for 15 min at 0°C. Excess NEM was removed by washing parasites two times with ice-cold potassium acetate buffer. Following NEM treatment, both nonpermeabilized and permeabilized parasites (treated with or without NEM) were suspended in potassium acetate buffer with or without ARS (2 mM MgATP was used in place of Na ϩ ATP in the ARS mixture) and incubated at 37°C for secretion to occur.
The effect of NSF on DG secretion was determined by incubating SLO-permeabilized parasites with 10 g/ml recombinant hamster NSF, or with the catalytically inactive N domain of NSF (10 g/ml) at 0°C for 1 h before switching to 37°C for secretion to occur. Prolonged incubation of parasites at 0°C was used due to the large size of full-length NSF (76 kDa). These experiments were conducted only in the presence of ARS, since NSF is unstable in the absence of ATP. Two mM Mg 2ϩ ATP was used in place of Na ϩ ATP in the ARS mixture.
The involvement of ␣-SNAP and Rab-GDI on DG secretion was determined by preincubating permeabilized parasites for 15 min at 15°C with various concentrations of ␣-SNAP (0.5 to 5 g/ml) and Rab-GDI (4 g/ml) in the presence or absence of ARS followed by switching to 37°C for secretion to occur.
DG secretion was also tested by preincubating permeabilized parasites with a 100 M final concentration of GTP␥S, ATP␥S, cAMP, and cGMP in the presence or absence of ARS at 0°C for 15 min, after which parasites were switched to 37°C for secretion to occur.
All results were expressed as percentage of BLA release, with 100% being the amount of BLA released by nonpermeabilized control parasites.

Statistical Analysis
Statistical analysis of the data was performed using InStat software for MacIntosh. Student t test was used for comparison of two groups. Results were considered significant at p values of Ͻ0.05.

Parasites Spontaneously Secrete Dense Granule Proteins in an Energy-and Temperature-dependent Fashion
Extracellular parasites spontaneously secreted both GRA3 and BLA in buffer. When compared with protein in total cell lysates, 11-14% of parasite GRA3 (Fig. 1, A and B) and 15-18% of BLA (Fig. 2B) were secreted during a 1-h incubation at 37°C. Secretion was abolished at 4°C or 15°C (Figs. 1A and 2A) or by the addition of 10 mM sodium azide (Fig. 1A) and was neither due to parasite lysis or to a general secretory response of extracellular parasites (Fig. 1A). No inhibition of BLA secretion was observed in cycloheximide-treated parasites ( Fig. 2A), a treatment that rapidly and completely inhibits protein synthesis (41). This latter result suggests that BLA secretion is occurring through preformed organelles, confirming previous observations (31).
We therefore asked whether conditions could be identified that might trigger or inhibit DG secretion. Based on suggestions that there is a burst of DG release after parasite entry (3,6), we reasoned that intracellular buffer conditions might augment release in comparison with extracellular buffer conditions. No difference in GRA3 or BLA release was observed when parasites were incubated in high sodium (HBSS) versus high potassium (AISS) buffer, except at later time points in the incubation for GRA3 (Fig. 1A). Secretion was not influenced by adding extracellular calcium or by chelating calcium with EGTA (Table I), although chelation of intracellular calcium with MAPTA-AM resulted in some inhibition of DG secretion. The addition of ATP, ADP, or AMP in the buffer had no effect on GRA3 or BLA release, nor was secretion altered by the addition of 10 mM glutathione at varying reduced:oxidized ratios. In contrast to one previous report (10), the addition of heat-inactivated fetal bovine serum in AISS buffer did not further enhance GRA3 or BLA release (Table I). In summary, these results suggest that DG release by extracellular T. gondii occurs constitutively.

Dense Granule Release in Permeabilized Parasites Is ATP-dependent
To facilitate the identification of signaling molecule(s) and protein machinery involved in DG secretion, we established a permeabilized cell secretion assay with T. gondii using streptolysin O as permeabilizing agent. Under the conditions developed for permeabilization, 50 -60% of parasites were permeabilized by 1 unit/ml of SLO (assessed by propidium iodide staining of the nucleus; not shown). Approximately one-quarter (see legend to Fig. 3A for calculation) of the cytosolic marker, green fluorescent protein, was released from permeabilized transgenic parasites (34) (Fig. 3A), showing that macromolecules are depleted.
Secretion of BLA was inhibited by 49 Ϯ 19% in SLO-permeabilized parasites as compared with untreated controls (Fig.  3B). The extent of BLA release during permeabilization was no greater than in controls incubated at 37°C without SLO. The addition of an ARS restored secretion to 81 Ϯ 20% of control. ARS had no effect on BLA secretion from nonpermeabilized parasites, further confirming our results (Table I). Preincubation of permeabilized parasites with an ATP depletion system (195 units/ml of hexokinase 10 mM D-glucose at 15°C for 30 min) resulted in a further 15-20% inhibition of secretion in the absence of ATP, in comparison with permeabilized parasites not treated with hexokinase and D-glucose (data not shown). This is likely to be the base line for secretion in the absence of ATP, since only 40 -60% of parasites were permeabilized by SLO. Altogether, these results show that ATP is necessary for DG secretion.

Neither Calcium nor cAMP Triggers DG Secretion from Permeabilized Parasites
An increase in intracellular calcium serves as a signal for secretion of dense matrix granules in many cells (42). In T. gondii, however, the addition of up to 100 M calcium to SLOpermeabilized parasites had no effect on BLA secretion in either the presence or absence of ARS. Chelation of intracellu-

FIG. 1. Kinetics of GRA3 secretion from extracellular T. gondii.
A, ELISA. The percentage of total GRA3 released to the supernatant was calculated from a standard curve, as described under "Materials and Methods." Secretion was similar in both high potassium (AISS, E) and high sodium (HBSS, f) buffer except at later points, where GRA3 release was significantly higher in AISS (p Յ 0.05) compared with HBSS. Secretion was energy-dependent (E) and inhibited at 4°C (small filled triangle) and 15°C (ϫ). Release of ROP2 (large filled triangle; a marker for rhoptry secretion) and of ␤-galactosidase (ϩ) (a transgenic cytoplasmic marker for parasite lysis) were less then 1-2% over 60 min. of incubation. Results represent mean Ϯ S.D. from four experiments performed in triplicate. B, parasites were incubated in AISS buffer, and the supernatant was collected, concentrated, and analyzed by SDS-PAGE immunoblot (top) as described under "Materials and Methods." Supernatants are from the equivalent of 10 8 parasites, whereas the pellet contains the equivalent of 10 7 parasites. The kinetics and extent of release of GRA3, determined by densitometric scanning (bottom panel), were similar to those determined by ELISA in A. Results shown are from a representative experiment performed twice.

FIG. 2. Kinetics of BLA secretion from extracellular T. gondii.
A, ELISA. BLA secretion was temperature-dependent (f, 4°C; E, 15°C; q, 37°C) but not dependent upon protein synthesis over the course of the assay (‚ ϭ cycloheximide). Results represent mean Ϯ S.D. of two experiments done in triplicate. Since the fraction of total BLA secreted could not be accurately determined by enzymatic assay (see "Materials and Methods"), the amount of BLA protein released was quantitated by immunoblot (B). B, immunoblot. Top, detection of BLA in the supernatant and in the pellet from 5 ϫ 10 6 parasites by SDS-PAGE and immunoblot, following incubation of parasites at 37°C in AISS buffer. Bottom, the percentage of BLA released was calculated by densitometric scanning of the immunoblot. 15-18% of total parasite BLA was secreted by extracellular parasites after 1 h at 37°C lar calcium with EGTA also had no effect on DG secretion (Fig.  4). MAPTA-AM, a more effective calcium chelator, inhibited BLA secretion by approximately 18% (Table II), a value somewhat lower than the inhibition by MAPTA-AM in nonpermeabilized extracellular parasites (Table I). Since cAMP can trigger granule exocytosis in the absence of calcium in some cells (43)(44)(45), we added cAMP to SLO-permeabilized parasites. As shown in Table II, cAMP did not alter BLA release. In combination with the data presented in Table I, these results further argue that DG release in T. gondii is constitutive.

The Involvement of NSF, ␣-SNAP, and GTP-binding Proteins in DG Exocytosis in T. gondii
We next explored the protein machinery mediating DG release in T. gondii. In other eukaryotic systems, NSF, the SNAP-SNARE machinery, and the Rabs have been shown to play important roles in exocytosis of regulated secretory organelles (14,17,18,(21)(22)(23)(24)(25).
Involvement of NSF-Treatment with 1 mM of the alkylating agent, NEM, resulted in 80 Ϯ 4% inhibition of BLA secretion from nonpermeabilized parasites and 65 Ϯ 3% inhibition of BLA secretion from permeabilized parasites in the presence of ARS (Fig. 5A). Although use of NEM at 1 mM has been previously used to inactivate NSF, NEM may have other effects. We therefore tested the effects of adding purified recombinant hamster NSF to permeabilized parasites. When parasites were incubated with recombinant hamster NSF and ARS, DG re-TABLE I DG secretion is not altered by extracellular versus intracellular buffer conditions Following liberation from host cells, nonpermeabilized parasites were incubated under conditions reflecting an extracellular environment (high Na ϩ , low K ϩ , high Ca 2ϩ , oxidizing, low ATP; HBSS) and conditions reflecting an intracellular environment (low Na ϩ , high K ϩ , low Ca 2ϩ , reducing, high ATP; AISS) or with other conditions, including added ARS, ATP␥S, GTP␥SADP, AMP, cAMP, cGMP, and heat-inactivated fetal bovine serum (FBS). The influence of extracellular calcium on DG secretion was further tested by either adding exogenous Ca 2ϩ (2 mM) to AISS buffer or by chelating Ca 2ϩ with either 5 mM EGTA or 500 M MAPTA-AM in AISS buffer. The influence of the oxidation/reduction state was tested by adding 10 mM glutathione at reduced to oxidized ratios of 60:1, 30:1, and 1:1 to AISS buffer. The released DG proteins were measured at 0 and 60 min postincubation by ELISA for GRA3 and by enzyme assay for BLA. The results are expressed as percentage of release and calculated with 100% being the amount of DG proteins released in AISS buffer. In most cases, there was no significant alteration in release of either GRA3 or BLA in comparison with AISS buffer. MAPTA-AM inhibited secretion of both BLA and GRA3 by 35 Ϯ 5% and 34 Ϯ 3%, respectively (*, p Յ 0.05). DG secretion was also significantly lower (**, p Ͻ 0.01) in the presence of vitamins and amino acids. The results of GRA3 release in HBSS are shown in Fig. 1 and therefore not included here. Unlike DG secretion, release of the rhoptry protein ROP2 from nonpermeabilized parasites was negligible (Ͻ1%) in any of the buffer conditions tested (not shown). 102 Ϯ 2 103 a ϩ glutathione (10 mM, 1:1) 100 a 102 a ϩ glutathione (10 mM) ϩ ATP (2 mM) 9 6Ϯ 3 9 9Ϯ 12 ϩ glutathione (10 mM) ϩ ADP (2 mM) 107 Ϯ 6 9 9Ϯ 5 ϩ glutathione (10 mM) ϩ AMP (2 mM) 9 5Ϯ 4 111 Ϯ 2 ϩ cAMP (2 mM) 9 2 Ϯ 2 8 9Ϯ 2 ϩ cGMP (2 mM) 9 9 Ϯ 2 104 Ϯ 2 ϩ CaCl 2 (0.1 mM) 9 8 Ϯ 2 9 9Ϯ 1 ϩ EGTA (5 mM) 9 3 Ϯ 4 100 Ϯ 2 ϩ MAPTA-AM (500 M) 6 9 Ϯ 3* 65 Ϯ 8* ϩ FBS (10%, HI) ND 99 Ϯ 4 ϩ vitamins ϩ amino acids 68 Ϯ 2** 63 Ϯ 2** HBSS ND 97 Ϯ 2 a The data represent the mean Ϯ S.D. of four experiments done in triplicate except in a few cases where only a single experiment was performed.
b ND, not determined.
FIG. 3. Extracellular parasites permeabilized with SLO release cytosolic markers and secrete less BLA. A, cytosolic GFP is released from SLO-permeabilized parasites. Parasites expressing cytosolic GFP were permeabilized with SLO for 2 min at 37°C and then incubated at 15°C for 1 h to deplete cytosol. Permeabilized parasites appeared phase-dense (arrows) and showed faint GFP fluorescence (arrows). Nonpermeabilized parasites were more refractile (phase) and demonstrated bright GFP fluorescence (FITC). As determined by fluorimetry, 15 Ϯ 3% of cytosolic GFP was released from SLO-permeabilized parasites after 60 min of incubation at 15°C. The average percentage of permeabilized parasites in these two experiments was 51 Ϯ 5% as judged by propidium iodide staining of the nucleus. Therefore, GFP release from the subpopulation of permeabilized organisms was calculated to be 27 Ϯ 2% ((percentage of GFP released from permeabilized parasites/percentage of permeabilized parasites) ϫ 100). Of note, an initial treatment of 37°C was required for efficient permeabilization with SLO; incubation at 4°C yielded only 10% permeabilization even when cells were shifted to 37°C after unbound SLO was washed away (not shown). B, inhibition and restoration of BLA secretion in SLO permeabilized parasites. BLA-expressing parasites were incubated with (ϩSLO) or without (control) 1 unit/ml SLO for 10 min at 4°C, followed by incubation at 37°C for 2 min. Supernatant was collected (Permeabilization step), and then an ARS was added to parasites, and incubation was continued at 37°C for an additional 30 min. Released BLA in the supernatant was determined by ELISA. Results shown are means Ϯ S.D. for seven experiments done in triplicate. *, p Յ 0.01 compared with control nonpermeabilized parasites; **, p Յ 0.01 compared with permeabilized parasites in the absence of ARS. lease was augmented in comparison with ARS alone (Fig. 5B). In contrast, incubation of parasites with the N domain of NSF, a fragment devoid of ATPase activity, did not augment BLA secretion from permeabilized parasites, arguing for the specificity of the NSF effect (Fig. 5B).
Involvement of ␣-SNAP-Treatment of permeabilized parasites with recombinant bovine ␣-SNAP resulted in stimulation of BLA secretion. Maximum secretion occurred at 2 g/ml of ␣-SNAP, in the presence of ARS (Fig. 6A). Based on this doseresponse curve, 2 g/ml ␣-SNAP was used in subsequent experiments. Secretion was augmented by ␣-SNAP in the presence of ARS (Fig. 6B). ␣-SNAP alone, in the absence of ARS, had no effect on DG secretion. These results suggest that DG secretion is regulated by ␣-SNAP and that heterologous ␣-SNAP proteins can interact with the SNARE/SNAP machinery to control DG docking and/or fusion.
Involvement of GTP-binding Proteins-Finally, we tested the involvement of GTP-binding proteins in DG secretion. The addition of 100 M GTP␥S resulted in significant enhancement of BLA secretion (38 Ϯ 4%) from permeabilized parasites in the presence of ARS (Table II), suggesting that GTP-binding proteins regulate DG exocytosis in T. gondii. To specifically explore the involvement of Rabs in DG secretion, SLO-permeabilized parasites were incubated with 4 g/ml recombinant bovine Rab-GDI. This resulted in 11 Ϯ 4% inhibition of DG secretion (data not shown). This difference, while small, was reproducible and is similar to the levels reported in other systems. DISCUSSION The molecular machinery controlling secretion of dense core granules is broadly conserved evolutionarily. Nonetheless, un- A, NEM inhibits DG secretion from permeabilized as well as nonpermeabilized parasites. SLO-permeabilized or nonpermeabilized parasites were preincubated with 1 mM NEM at 0°C for 15 min. Following incubation, parasites were washed, resuspended in fresh buffer, and incubated for 30 min at 37°C. Released BLA was determined by an enzymatic assay and expressed relative to control parasites not incubated with NEM (100%). N-Ethylmaleimide significantly inhibited BLA secretion from both nonpermeabilized (*, p Ͻ 0.01 compared with untreated nonpermeabilized parasites) and from permeabilized parasites in the presence of ARS (**, p Ͻ 0.01 compared with secretion from permeabilized parasites with ARS alone). The data are means Ϯ S.D. from three different experiments done in triplicate. B, recombinant hamster NSF stimulates DG secretion in SLO-permeabilized T. gondii. Permeabilized parasites were preincubated with 10 g/ml of hamster-NSF at 4°C for 1 h followed by incubation at 37°C for 30 min. NSF augmented BLA secretion by 30 Ϯ 11%. The addition of 10 g/ml of the catalytically inactive N domain of NSF (N-NSF) did not enhance BLA secretion. The results represent the mean Ϯ S.D. of four experiments done in triplicate. *, p Ͻ 0.05 compared with permeabilized parasites with ARS alone. til this study, there has been no direct evidence that this machinery functions in pathogenic unicellular protozoan parasites, either for constitutive or regulated secretion. In contrast to this conserved aspect of dense granule release, secretion of T. gondii dense granules is calcium-independent, reflecting the unusual nature of the protein sorting machinery within the organism (31).
In most systems, secretion from dense matrix granules is regulated by changes in intracellular calcium concentration (11), while constitutive transport between the trans-Golgi network and plasma membrane is calcium-independent (46,47). Preliminary experiments indicated that treatment of extracellular parasites with (i) the calcium ionophore A23187, (ii) the ER CaATPase inhibitor thapsigargin, or (iii) NH 4 Cl, (under conditions reported to increase intracellular calcium levels in T. gondii; Ref. 48) had no effects on secretion of GRA3 and BLA (data not shown). Theoretical considerations also support the notion that increased intracellular calcium is unlikely to be the trigger for continuous exocytosis of dense granules from intracellular T. gondii, since the parasite resides in a low calcium environment and would not be capable of rapidly replenishing intracellular calcium by opening plasma membrane calcium channels (reviewed in Ref. 49). In order to examine this ques-tion further and more precisely control intracellular calcium levels, we developed a permeabilized cell system and unequivocally demonstrated that increases in intracellular calcium do not trigger DG release.
Spontaneous release of "regulated proteins" in the absence of a stimulus is generally ascribed to either low level release of dense core granules or simultaneous delivery of the regulated proteins to both regulated and constitutive secretory vesicles (50,51). The distinction between basal and constitutive release of dense core vesicles depends on the capacity to identify triggers (typically a rise in intracellular Ca 2ϩ ) for granule release in the former instance (46). The results shown above demonstrate that DG release in T. gondii is not triggered by an increase in intracellular Ca 2ϩ , precluding the use of this signal to make the distinction between basal and constitutive release. Based on our previous results with brefeldin A (31), showing quantitative delivery of GRA3 and BLA to DG, and results here showing that short term treatment of extracellular parasites with cycloheximide had no effect on DG release, it is unlikely that DG proteins are simultaneously delivered to DG and constitutive secretory vesicles. An alternative possibility is that the DG in Toxoplasma are analogous to the immature secretory granules in endocrine and exocrine cells (52)(53)(54)(55)(56)(57)(58) and that secretion occurs via budding of small constitutive secretory vesicles from the DG. Vesicle budding from immature secretory granules in mammalian cells is brefeldin A-sensitive (59, 60), arguing against this explanation for constitutive secretion of dense granule proteins in T. gondii. Moreover, neither we nor others have observed structures compatible with immature secretory granules in the organism. Taken in total, the results are most consistent with the notion that DG secretion is constitutive in T. gondii.
The discovery that hamster NSF and bovine ␣-SNAP augment DG secretion provides the first functional demonstration that the NSF/SNAP/SNARE system operates in a pathogenic protozoan parasite. Although we do not yet have direct evidence for the involvement of endogenous T. gondii components, the function of hamster and bovine proteins in the parasite system suggests that the T. gondii homologous will be structurally similar with their mammalian counterparts (reviewed in Ref. 27).
Numerous Rabs have been identified in Toxoplasma (61), but the specific molecules involved in regulating DG secretion are not known. Since Rab-GDI exhibits broad substrate specificity, even across species barriers (reviewed in Refs. 62 and 63), the activity of bovine protein in T. gondii does not reveal which isoform of Rab is involved. Other experiments in our laboratory suggest that both T. gondii Rab6 and Rab11 are involved in DG secretion. 2 In light of the inability to demonstrate calciumregulated secretion in Toxoplasma, however, it is interesting to note that Rab3, commonly associated with calcium-dependent exocytosis (25,64), has not been identified in this parasite.
Treatment of permeabilized parasites with the proteolytically active light chain of tetanus toxin, which cleaves synaptobrevin/VAMP (65) and inhibits both calcium-dependent (66) and calcium-independent (67) granule secretion, had no effect on BLA release (data not shown). Unfortunately, interpretation of this experiment must await identification of T. gondii VAMP homolog, because not all VAMP/synaptobrevins are cleaved by TeTx (66).
The molecular basis for the GTP␥S-mediated enhancement of DG secretion is unknown. Calcium-independent GTP␥S-augmented secretion is also observed in mast cells and neutrophils (68,69), where it is mediated by ARF-1 activation of phospho-2 T. Stedman and K. A. Joiner, unpublished observations. lipase D to generate phosphatidylinositol bisphosphate (70). This is the most likely explanation for the GTP␥S-mediated enhancement of DG exocytosis in T. gondii, where GTP␥S is revealing the components involved in the constitutively occurring process. Alternatively, it is possible that a parasite Gprotein-coupled plasma membrane receptor could be linked to DG exocytosis. Heterotrimeric GTP-binding proteins have been identified in T. gondii, and AlF is reported to increase the extrusion of filamentous structures from the parasite surface (71), although in our hands BLA release was altered by NaF alone, making interpretation of these results difficult. By identifying the mechanism for the GTP␥S-augmented secretion, the possible contribution of a regulated component to DG exocytosis (3,6) in T. gondii may also be defined.