T lymphocytes export proteasomes by way of microparticles: a possible mechanism for generation of extracellular proteasomes

The 20S proteasome is almost exclusively localized within cells. High levels of extracellular proteasomes are also found circulating in the blood plasma of patients suffering from a variety of inflammatory, autoimmune and neoplastic diseases. However, the origin of these proteasomes remained enigmatic. Since the proteome of microparticles, small membrane enclosed vesicles released from cells, was shown to contain proteasomal subunits, we studied whether intact proteasomes are actively released into the extracellular space. Using human primary T lymphocytes stimulated with CaCl2 and the calcium ionophore A23187 to induce membrane blebbing we demonstrate that microparticles contain proteolytically active 20S proteasomes as well as the proteasome activator PA28 and subunits of the 19S proteasome regulator. Furthermore, our experiments reveal that incubation of in vitro generated T lymphocyte-microparticles with sphingomyelinase results in the hydrolysis of the microparticle membranes and subsequent release of proteasomes from the vesicles. Thus, we here show for the first time that functional proteasomes can be exported from activated immune cells by way of microparticles, the dissolution of which may finally lead to the generation of extracellular proteasomes.


Introduction
Extracellular proteasomes circulating in blood plasma (circulating proteasomes) are potential diagnostic biomarkers for various malignancies like systemic lupus erythematodes (SLE) [1], melanoma [2], multiple myeloma [3], epithelial ovarian cancer [4] and hepatocellular carcinoma [5,6]. Clear correlations were found to exist between the plasma level of extracellular proteasomes and the malignant transformation of cirrhotic liver cells [6] as well as residual tumour mass after therapy in patients suffering from epithelial ovarian cancer [4].
The proteasome catalyses the degradation of intracellular proteins and is involved in vital functions like cell cycle progression, differentiation, immune defence and the stress response [7,8]. The barrel-shaped 20S proteasome is composed of four stacked, sevenmembered rings, two inner b and two outer a rings. The a rings provide contact to 19S regulators or to the proteasome activator PA28 thus forming 26S proteasomes (19S-20S), 30S proteasomes (19S-20S-19S), PA28-20S proteasome-complexes or hybrid proteasomes (PA28-20S-19S), respectively. The b rings contain the three proteolytically active subunits with b1 exhibiting caspase-like, b2 trypsin-like and b5 chymotrypsin-like activity. Cytokines like interferon-c induce the expression of the immuno-subunits b1i (LMP2), b2i (MECL1) and b5i (LMP7), which replace the standard-subunits b1, b2 and b5 resulting in the biogenesis of immuno-proteasomes [7,8]. Immunoproteasomes are constitutively generated by cells involved in the immune response like spleenocytes, thymocytes, dendritic cells and others [9].
Circulating proteasomes from human plasma are proteolytically active 20S proteasomes [10]. They were also detected in the lung epithelial lining fluid of patients with acute respiratory distress syndrome [11] and in cerebrospinal fluid [12]. Apart from their importance as diagnostic biomarkers extracellular proteasomes may also exert various extracellular functions [13,14]. However, so far the mechanisms by which proteasomes are released into the extracellular space remained obscure [15]. Although a passive efflux of proteasomes as a result of cell death occurring in the course of various diseases cannot be excluded, several observations argue in favour of a specifically directed mechanism [13,15].
Proteasomal subunits contain no signal sequence for export via the classic secretory pathway. Therefore, alternative secretion mechanisms might be responsible for proteasome release to the extracellular environment [16][17][18], e.g. secretory lysosomes and plasma-membrane-residing transporters. But so far there exists no evidence for a proteasome-interacting transporter at the plasma membrane. Another alternative export pathway releases exosomes, which are 50-100 nm vesicular structures. Finally, a membrane blebbing pathway was described whereby cytoplasmic content is entrapped in protrusions of the membrane, which are then directly shed as vesicles (microparticles) of heterogeneous size (0.1-1 lm) from the plasma membrane of activated cells. During the last decade evidence was provided that microparticles (MP) can act as messengers between cells since they transport a variety of molecules [19][20][21][22]. Proteomic analyses of microparticles derived from platelets and a T lymphocyte cell line have revealed the presence of subunits of the proteasome, the proteasome activator PA28 and the ubiquitination machinery [23,24]. By means of immunoelectron microscopy proteasomes were also identified in vesicles in cerebrospinal fluid [12]. Therefore, we investigated whether membrane blebbing and shedding of vesicles as well as their subsequent breakdown may be a mechanism by which proteasomes are released from cells thereby generating extracellular proteasomes.

Cell culture
T lymphocytes were isolated from buffy coats of healthy donors provided by German Red Cross (Berlin, Germany). They were maintained in RPMI 1640 medium without phenolred supplemented with 10% (v/v) heat-inactivated FCS, 100 U/100 lg/ml penicillin/streptomycin and 2 mM L-glutamine (complete medium) at 37°C in a 5% CO 2 atmosphere up to 1 week after isolation. For the experiments 5 9 10 7 cells were centrifuged (300 9 g, 8 min.) and taken up in 6 ml of serum-free or microparticle-free (MP-free) medium. MP-free medium is complete medium supplemented with FCS centrifuged at 100,000 9 g over night to remove bovine plasma microparticles.

Isolation of T lymphocytes
Buffy coats were diluted with 0.5-1 volume of RPMI 1640 medium (without supplements) and 40 ml of the cell suspension were covered on 12 ml Biocoll reagent and centrifuged for 30 min. at 20°C (675 9 g). Peripheral blood mononuclear cells (PBMCs) remaining in the layer above the Biocoll reagent were isolated and washed three times in RPMI (8 min., 300 9 g, 4°C). For separation of monocytes from T and B cells, PBMCs were diluted in RPMI and incubated in cell culture flasks for 3-4 hrs at 37°C in a 5% CO 2 atmosphere. Under these conditions monocytes adhere to the cell culture flask and the supernatant contains the T and B lymphocytes. 85% of the isolated cells were CD3-positive T cells as analysed by flow cytometry (data not shown).
Preparation of microparticles 5 9 10 7 T cells were stimulated with 2 lM A23187 calcium ionophore in serum-free or MP-free medium supplemented with 1 mM CaCl 2 [26]. Control cells were stimulated with DMSO or DMSO plus 1 mM CaCl 2 at 37°C in a 5% CO 2 atmosphere for 45 min., respectively . The medium containing cells and microparticles was subjected to differential centrifugation at 470 9 g for 5 min. at 4°C to remove the cells. The supernatant was transferred to a second centrifugation (1400 9 g, 10 min., 4°C) to remove larger debris. To isolate microparticles, the supernatant was centrifuged at 100,000 9 g for 20 min. at 4°C. The centrifugation step was repeated two times if washing of the microparticles with PBS was required. The effect of Ca 2+ /A23187 treatment on cell viability was checked by trypan blue staining and LDH release (CytoTox ONE) according to the manufacturer's protocol.

Quantification of 20S proteasome
Quantification of 20S proteasome was carried out by ELISA after breakdown of the microparticles in PBS supplemented with 1% Tween 20 by repeated freeze-thaw cycles and subsequent centrifugation to remove residual lipids (45 min., 20,000 9 g, 4°C). The capturing antibody against the proteasome-subunit a6, dissolved 1 lg/ml PBS, was bound to an ELISA plate over night at 4°C. After washing with PBST (PBS containing 0.2% Tween 20), the plate was blocked with 1% albumin fraction V dissolved in PBST for 2 hrs and the washing step was repeated. On each ELISA plate purified erythrocyte 20S proteasome was used as calibration standard. The samples (plasma samples were diluted 1:2) and proteasome standards were detected with a polyclonal antibody to 20S proteasome (dilution 1:1000), the binding of which was photometrically quantified with a peroxidase-coupled rabbit IgG antibody and TMB substrate (BD Biosciences). The detection range was 2-60 ng of proteasome.

Flow cytometric detection and quantification of microparticles
Microparticles were generated and isolated as described and washed two times in sterile filtered PBS. After the last centrifugation step, the microparticles were suspended in 200 ll sterile-filtered annexin V binding buffer, transferred to Trucount â tubes and stained with FITC-annexin V. After 15 min. of incubation the samples were diluted by 400 ll annexin V binding buffer and directly analysed with an FACS Calibur flow cytometer (BD Biosciences). The size analysis with Megamix beads was accomplished according to the manufacturer′s protocol (Biocytex, Marseille, France).

Western blot analysis
T cells (5 9 10 7 ) were stimulated with 1 mM CaCl 2 and 2 lM A23187 calcium ionophore in serum-free medium for 45 min. before cells and microparticles were separated by centrifugation. The pelleted cells were washed with PBS and resuspended in 100 ll of RIPA buffer (10 mM Tris-HCl pH 7.5, 100 mM NaCl, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS) supplemented with Complete protease inhibitor mix (Roche, Basel, Switzerland); the microparticles were suspended in 20-40 ll RIPA buffer. Cells and microvesicles were then disrupted by repeated freeze-thaw cycles and undissolved material precipitated at 20,000 9 g for 40 min. at 4°C. The microparticle supernatants were mixed with sample buffer (20 mM Tris-HCl pH 6.8, 10% (w/v) glycerol, 5% b-mercaptoethanol, 2% SDS, 0.05% bromphenol blue) and entirely loaded on the SDS-gel. Protein concentrations were measured in cell lysate by means of BCA assay and equal amounts of protein were subjected to SDS-PAGE (12% w/v). The proteins were immunoblotted and detected by using the ECl Reagent.

Digestion of microparticles with sphingomyelinase
For enzymatic breakdown of microparticles, the isolated microvesicles were resuspended in 10 mM HEPES, pH 7.4, containing 140 mM NaCl, 2.7 mM KCl, 1 mM MgCl 2 , 5 mM glucose and 1 mM CaCl 2 and then incubated at 37°C with or without 0.1 U/ml sphingomyelinase from Bacillus cereus [26]. After 1 hr of enzymatic digestion remaining vesicle residues were pelleted at 20,000 9 g for 40 min. and the supernatants analysed for 20S proteasomes by ELISA. The membrane fraction was resuspended in RIPA buffer supplemented with complete protease inhibitor cocktail, the membranes were disrupted by repeated freezethaw cycles and after centrifugation the supernatant was subjected to SDS-PAGE.

Blood collection and microparticle isolation of human plasma
Twenty millilitre peripheral blood was collected from healthy donors with 21-gauge needles in Vacutainer tubes (BD Biosciences) supplemented with citrate. Platelet-free plasma was produced by different centrifugation steps, 500 9 g for 10 min and two times at 2000 9 g for 30 min. [27]. One millilitre platelet-free plasma was diluted 1:4 (v/v) with PBS and microparticles were isolated at 100,000 9 g for 40 min. at 4°C, washed with PBS and collected at 100,000 9 g for 20 min. The supernatant fraction was used for the quantification of free plasma proteasomes and the pelleted fraction for vesicle-bound proteasomes. The latter were also analysed by immuno blotting.

Statistical analysis
Statistical comparison of two groups of data was done by Student's ttest or in case of equal variance by Mann-Whitney Rank Sum Test. Error bars display standard error of the mean (AESEM) values. P-values lower than 0.05 were considered statistically significant.

Cellular and vesicular T lymphocyte proteasomes
Microparticles of T lymphocytes and other blood cells were found in blood plasma of healthy individuals and patients suffering from various diseases [28]. We hypothesized that microvesicles may be a source for extracellular proteasomes. To allow comparison of T cell derived proteasomes with those that may be present in T cell derived microvesicles we initially analysed proteasomes in primary T lymphocytes of healthy donors. T cell lysates incubated with proteasomespecific fluorogenic peptide substrates showed all three proteasomal activities. The chymotrypsin-and trypsin-like activities could be inhibited by epoxomycin (P ≤ 0.001 and 0.002, respectively, Fig. 1A). To  investigate, which type of proteasome is present in T lymphocytes, lysates of T cells were subjected to SDS-PAGE and analysed for standard-(b1, b2 and b5) and immuno-subunits b1i (LMP2), b5i (LMP7). T cell lysates showed signals solely of immuno-subunits LMP2 and LMP7 (Fig. 1B) indicating that exclusively immunoproteasomes are constitutively present in these cells. In addition, the T cell lysates contained components of the 19S regulatory particle (Rpt3) and the proteasome activator PA28 (PA28a and b) (Fig. 1C). To explore whether the culture medium of T lymphocytes contains extracellular proteasomes the culture medium was subjected to differential centrifugation to separate remaining cells and debris. Thereafter, the cell-free medium was ultracentrifuged to precipitate microvesicles and to obtain a vesicle-free culture supernatant. Both the supernatant and the vesicle fraction were incubated with the membrane-permeable substrate Suc-LLVY-AMC to measure proteasomal chymotrypsin-like activity. As shown in Figure 1D proteasomal activity was measured only in the microvesicle fraction indicating that extracellular proteasomes in the cell culture medium are associated with vesicular structures.

Characterization of microparticles shed by activated T lymphocytes
If vesicle shedding by T cells results from cell membrane blebbing, microvesicles are expected to contain proteins originating from their parental cell [28]. Therefore, the vesicular membranes were dissolved by detergents and tested for the presence of a T cell membrane-localized protein, the leucocyte-derived integrine lymphocyte-function associated antigen-1 (LFA-1). Figure 1E clearly revealed the existence of LFA-1 in the vesicle membranes suggesting that membrane blebbing and shedding occurs continuously.
From earlier studies it is known that cells can be activated by the A23187 calcium ionophore and CaCl 2 resulting in an enhanced segregation of microparticles from the plasma membrane [26]. Therefore, we stimulated T lymphocytes with A23187 and CaCl 2 before the vesicle fraction was isolated and analysed by flow cytometry (Fig. 2). During the shedding process, the parental cells lose their membrane symmetry at vesicle exit sites. This leads to exposure of phosphatidylserine residues to the outer leaflet of the plasma membrane [28] thereby allowing the specific binding of annexin V to phosphatidylserine. Together with the particle size this feature can serve as a marker to differentiate microparticles from other particulate structures [29]. As shown in Figure 2A flow cytometric analysis of the vesicles derived from the T cell fraction revealed binding of annexin V-FITC and this signal was strongly enhanced after stimulation of vesicle blebbing. A comparative dot plot of lymphocyte membrane vesicles and fluorescent Megamix beads affirmed that the annexin V-positive vesicles were in the microparticle size gate (Fig. 2B). For quantification of the microparticle count, 5000 Trucount â beads were counted in a defined area and compared with the absolute number of microparticles in the microparticle gate. Non-stimulated lymphocytes showed a basic shedding of 4246 AE 1114 microparticles and this was significantly enhanced to 38,095 AE 9857 by the activation of T cells (P ≤ 0.001, Fig. 2C and D). The enrichment of the microparticle fraction could also be demonstrated by immunoblot analysis revealing the increased expression of LFA-1 (Fig. 2E). To exclude that any kind of cell death contributes to the Ca 2+ /A23187-induced release of microparticles we measured the amount of live and dead cells by trypan blue staining and also the release of LDH from T cells. As shown in Figure 2F and G there was no significant difference with regard to both parameters in the cells treated with and without Ca 2+ /A23187 stimulation.

Analysis of proteasomes of microparticles shed by T lymphocytes
To investigate whether the microparticles may be a source of extracellular proteasomes, we extracted their proteins by repeated freeze-thaw cycles. After centrifugation 20S proteasomes were detected in the supernatant fraction and quantified by ELISA (Fig. 3A). The microparticle fraction of resting 5 9 10 7 lympho-cytes contained 4 AE 1 ng 20S proteasomes, whereas the same amount of A23187/CaCl 2 treated T cells contained 28 AE 3 ng 20S proteasome (corresponding to 4246 AE 1114 versus 38,095 AE 9857 microparticles measured in Fig. 2D). To study whether the extracellular proteasomes are proteolytically active native gel electrophoresis was performed to separate the different proteasome complexes of microparticles and of T cell lysates. As a control we used purified human erythrocyte 26S proteasome.  Figure 3B shows the in-gel (substrate overlay technique) trypsinlike activity of proteasomal complexes after their separation by native gel electrophoresis. T cell lysates contain two proteolytically active complexes. As evidenced by immunoblotting with antibodies to subunit a6 (20S proteasome), to Rpt3 (19S regulator) and to PA28 (PA28b), the bigger of the two complexes contains the 19S regulator and PA28 in addition to the 20S proteasome. Thus, hybrid proteasome (19S-20S-PA28) is the predominant proteasome complex in T lymphocytes. The other proteolytically active complex in T cell lysates exhibits an electrophoretic mobility slightly slower than that of the 20S proteasome and represents 20S-PA28 proteasome complexes as shown by immune reactivity against the proteasome subunits a6 and PA28b. In contrast microparticles contain just proteolytically active 20S proteasome (Fig. 3C). Immunoblotting confirmed that proteasomes in microvesicles are immunoproteasomes containing the subunits b1i/LMP2 and b5i/LMP7. As reflected by an enhanced signal for b1i/LMP2, b5i/ LMP7 and LFA-1 activation of T cells by the calcium ionophore A23187 and CaCl 2 results in increased generation of microparticles (Figs 2E and 4A). On the other hand, inhibition of microparticle blebbing by EGTA resulted in a strong reduction of microparticle derived 20S proteasomes (Fig. 4C). Notably, the Ca/A23187 enhanced packaging of T cell 20S proteasomes into microparticles seems to go along with an export of proteasomal regulatory complexe subunits (Fig. 4B), which are not associated with the 20S core complex (compare Fig. 3B and C), because only under this experimental condition the 19S regulator subunit Rpt3 and PA28 were detectable.

Microvesicles from human blood plasma contain proteasomes
Since proteasomes are released from cells by microvesicle shedding, we were interested to know whether proteasome-containing microvesicles can also be isolated from blood plasma of human volunteers. In preceding experiments we had found that the concentration of circulating proteasomes in plasma was higher when the plasma probes were stored frozen before the ELISA was carried out (239 AE 138 ng/ml frozen versus 148 AE 132 ng/ml unfrozen). The proteasome concentrations measured in frozen and unfrozen platelet free plasma were 140 AE 118 ng/ml versus 140 AE 114 ng/ml, respectively. Therefore, platelets were removed from the plasma of six healthy donors before microvesicles were isolated by centrifugation. As shown by immunoblotting microvesicles of all individuals contained 20S proteasomes (Fig. S1). Their mean concentration was 14.8 AE 3.8 ng 20S proteasome/ml plasma. Analysis of the remaining supernatant fraction revealed that in all samples proteasomes were also present in a vesicle-free form, the mean concentration of which was 164 ng/ml plasma (range 80-475 ng/ml). Thus, the ratio of microparticle bound to free 20S proteasome is about 1:11 suggesting that in plasma a mechanism must exist allowing the release of proteasomes from microvesicular structures.

Microparticle breakdown by lipases releases free extracellular proteasomes
The levels of lipid-degrading enzymes such as secretory phospholipase A2 (sPLA2) and sphingomyelinase (SMase) are known to be increased in the blood plasma of patients suffering from several types of disorders [30][31][32][33]. Fourcade and co-workers reported that the combined action of sPLA2 and SMase released free fatty acids from vesicle membranes [26]. Therefore, we investigated whether of a commercially available SMase might hydrolyse microparticle membranes resulting in the release of proteasomes from microvesicles. Figure 5A shows that the amount of proteasome indicated by sub-unit a6 (detection of subunit LMP7 showed the same results; data not shown) and of the cell surface protein LFA-1 was significantly reduced in the vesicle fraction treated with SMase when compared to control samples. Accordingly, the proteasome content in the supernatant fractions of the SMase digestion samples was significantly increased and the liberation of proteasomes from vesicular structures by SMase appeared to be more efficient than repeated freeze-thaw cycles (compare Figs 5B and 3A). These data show that

Discussion
Since proteasomes were detected to circulate in blood plasma and their concentration is increased in patients with various malignancies, the question arose whether circulating proteasomes are released from damaged cells. Although initially a positive correlation between the elevated serum proteasome levels and a marker for cell death, like LDH, was observed in patients with malignant neoplasms and some autoimmune diseases [5,13] most other investigations in patients suffering from malignant and non-malignant diseases [11,[34][35][36] found only poor correlations between markers of cell lysis and the level of circulating proteasomes. Therefore, processes other than cell disruption may contribute to the level of circulating proteasomes and we show here that T lymphocytes traffic proteasomes into their environment packed in microparticles (Figs 1D, 4A and B). This mechanism is used for basal as well as for enhanced outward transfer, since a rise in proteasome-containing microparticle release was observed in the presence of Ca 2+ and a Ca-ionophore (Figs 2 and 4). Since many different stimuli for the activation of cells are responsible for microparticle release like the early phase of apoptosis, Ca 2+ -ions, phorbol esters, cytokines, complement factor C5a, C-reactive protein and high concentrations of ATP [16,37], it is not surprising that several diseases linked to a misguided immune response like SLE, polymyositis, primary Sj € ogren′s syndrome, sepsis and RA as well as hematologic malignancies and other forms of cancer go along with increased levels of serum microparticles and circulating proteasomes (Table S1).
Calculating the proteasome content in microparticles shed from cultured T cells (Figs 1 and 2) yields about 1 ng proteasome/1000-1300 microparticles. Since the number of microparticles in healthy individuals vary between 10 5 and 10 6 /ml blood [38], a quantitative release of their content into plasma would result in a concentration of 10-1000 ng proteasome/ml plasma. Considering that plasma microparticles are thought to have various cellular origins including platelets, erythrocytes, leukocytes, endothelial cells and vascular smooth muscle cells [37], 400 ng circulating proteasomes/ml plasma as measured in normal individuals [15] fits well to our experimental findings.
Since circulating proteasomes in patients were directly quantified by ELISA technique, they must exist in a free, non-vesicular form, allowing binding of antibodies for reproducible quantification. Applying an in-vitro assay it was shown that two phosphodiesterases, cPLA2 and SMase, catalyse microparticle membrane turnover [26] and thus may release the vesicle content including proteasomes into body fluids. Our experiments using a commercially available bacterial SMase led to pronounced microparticle degradation and to liberation of vesicular proteasomes (Fig. 5) and thus sustain the likelihood of this mechanism to be physiologically relevant. Proinflammatory mediators foster the release of lipid-degrading enzymes, as sPLA2 was found to be released from rat calvarial bone-forming cells by IL-1 and TNF-a influence [39]. Acid-SMase was reported to be secreted by endothelial cells by cytokine activation including interferon-c and IL-1b [40], suggesting a pathophysiological relevance for these phospholipases. Elevated plasma levels of lipid-hydrolyzing enzymes have been linked to various disorders like rheumatoid arthritis (sPLA2) and sepsis (sPLA2, acid-SMase) [31,41,42] that are also characterized by increased levels of extracellular proteasomes [15]. Since, the levels of microparticles were inversely correlated with sPLA2 activity [43], elevated levels of extracellular proteasomes are a reasonable consequence. Therefore, our in vitro experiments may indeed mimic the in vivo situation.
Our data are compatible with proteomic analyses of microparticles, which identified subunits of proteasomes and their regulators in platelet and T lymphocyte microvesicles [23,24]. Moreover, our data show that proteasomes released from microparticles are proteolytically active. Whether proteasomes exhibit any proteolytic function in the extracellular environment remains unknown and is a matter of debate [14,15]. Recently, Lai and colleagues detected that exosomes deriving from the endosomal compartment of mesenchymal stem cells also contain proteolytically active 20S proteasomes and they proposed that exosomal proteasomes may have a cardioprotective effect [44]. On the other hand, microparticles as well as exosomes may be a potent source for autoantigens and provoke an immune response, because autoantibodies to proteasomal subunits were detected in the plasma of individuals with autoimmune disorders [45,46].
In summary, the data presented here provide a novel link between microparticle shedding and extracellular proteasomes and will require further investigations on microparticle count, plasma phospholipase activity and level of circulating proteasomes in healthy and diseased volunteers to evaluate the impact of this system on disease activity and prognostic value.