Research ArticleAutoimmunity

The 20S proteasome core, active within apoptotic exosome-like vesicles, induces autoantibody production and accelerates rejection

See allHide authors and affiliations

Science Translational Medicine  16 Dec 2015:
Vol. 7, Issue 318, pp. 318ra200
DOI: 10.1126/scitranslmed.aac9816

Friendly fire from organ failure

Despite advances in organ transplantation, rejection still poses a substantial risk. Autoantibodies contribute to rejection, but how these autoantibodies are generated remains unknown. Dieudé et al. found that injection of apoptotic exosome-like vesicles apoExo stimulated autoantibody production in mice, which led to increased graft rejection after transplant. The apoExo contained active 20S proteasome core complexes, and inhibition of proteasome activity decreased the immunogenicity of apoExo and graft rejection in transplanted mice. Circulating apoExo and increased anti-autoantibody titers were also observed in mouse models of ischemia-reperfusion injury, suggesting that the same organ failure that necessitates the transplant might increase the risk of rejection. Therefore, proteasome inhibitors could provide a new therapeutic avenue for graft rejection.

Abstract

Autoantibodies to components of apoptotic cells, such as anti-perlecan antibodies, contribute to rejection in organ transplant recipients. However, mechanisms of immunization to apoptotic components remain largely uncharacterized. We used large-scale proteomics, with validation by electron microscopy and biochemical methods, to compare the protein profiles of apoptotic bodies and apoptotic exosome-like vesicles, smaller extracellular vesicles released by endothelial cells downstream of caspase-3 activation. We identified apoptotic exosome-like vesicles as a central trigger for production of anti-perlecan antibodies and acceleration of rejection. Unlike apoptotic bodies, apoptotic exosome-like vesicles triggered the production of anti-perlecan antibodies in naïve mice and enhanced anti-perlecan antibody production and allograft inflammation in mice transplanted with an MHC (major histocompatibility complex)–incompatible aortic graft. The 20S proteasome core was active within apoptotic exosome-like vesicles and controlled their immunogenic activity. Finally, we showed that proteasome activity in circulating exosome-like vesicles increased after vascular injury in mice. These findings open new avenues for predicting and controlling maladaptive humoral responses to apoptotic cell components that enhance the risk of rejection after transplantation.

INTRODUCTION

Antibodies targeting the vasculature of solid allografts are major risk factors of poor outcomes in solid organ transplantation (17). Although donor-specific antibodies (DSAs) to human leucocyte antigen (HLA) play a major role in allograft vascular injury, autoantibodies are increasingly recognized as important contributors to allograft vascular rejection and decreased long-term survival (819). Autoantibodies to vimentin, angiotensin II type 1 receptor, collagen V, tubulin, fibronectin, and perlecan/LG3 have been associated with increased rejection rates and reduced survival of heart, kidney, and lung allografts (8, 17, 2028). The classic view in the field was that autoantibody titers increase during acute rejection episodes secondary to the release of danger-associated molecular patterns (DAMPs). However, recent studies have reported the presence of autoantibodies before transplantation in de novo transplant patients, excluding rejection as a sine qua non factor for production of autoantibodies (8). Anti-LG3 and anti-angiotensin receptors have been detected in untransfused*untransplanted males and untransfused*untransplanted females who have never been pregnant (8, 29), also suggesting that allosensitizing events are not required for their production (8, 17, 2025).

Abnormal reactivity to apoptotic cells plays a role in fuelling the production of at least some of these autoantibodies. Two recent studies have identified circulating levels of immunoglobulin G (IgG) reactive with apoptotic cells as a marker of antibody-mediated rejection (ABMR) (30, 31) in renal transplant patients. The pretransplant detection of antibodies reactive with apoptotic cells predicted increased risk of ABMR and reduced long-term allograft survival. Our group also described an association between pretransplantation titers of IgG autoantibodies against the C-terminal perlecan fragment LG3 and the risk of acute vascular rejection in renal transplant patients (8). Apoptosis of endothelial cells (ECs) is a major pathway regulating perlecan proteolysis and production of LG3 (32, 33). The association between autoantibodies and apoptosis is intriguing, given the classical notion that apoptosis represents a nonimmunogenic, if not tolerogenic, type of regulated cell death (34, 35). Apoptotic cells are known to release apoptotic bodies, which, when engulfed by professional phagocytes, trigger the secretion of anti-inflammatory cytokines (for example, transforming growth factor–β and interleukin-10) (34, 36) and increase the number of regulatory T cells. Autoantibodies to components of apoptotic cells or apoptotic bodies have been identified in lupus patients (37) because of abnormalities in clearance of apoptotic cells that prevent phagocytosis-induced immune tolerance and lead to secondary necrosis (34). However, no propensity toward lupus or other autoimmune diseases was found in patients with autoantibodies against apoptotic cells or perlecan/LG3 (8, 30).

Recent work by our group and others has highlighted a major role for apoptosis and caspase-3 activation in controlling the release of extracellular vesicles (EVs) that appear to differ from classical apoptotic bodies (3840). Apoptotic bodies, classically ranging in size from 1 to 5 μm and characterized by the presence of histone components, are released by apoptotic cells through caspase-dependent pathways (41). Caspase activation can also lead to the release of smaller EVs, referred to as apoptotic exosome-like vesicles and ranging in size from 30 to 100 nm. The immune-modulating functions of apoptotic exosome-like vesicles are still largely uncharacterized.

Here, we aim to evaluate the immunogenic potential of the various extracellular structures released by apoptotic ECs downstream of caspase-3 activation in the early phases of apoptosis. We use large-scale proteomics to profile the protein signatures of apoptotic bodies and apoptotic exosome-like vesicles. In addition, we compare the capacity of apoptotic exosome-like vesicles and apoptotic bodies to trigger the production of autoantibodies in transplanted and nontransplanted mice and shape the severity of rejection.

RESULTS

Apoptotic ECs release different types of EVs downstream of caspase-3 activation

To characterize the various types of EVs released by apoptotic ECs, human umbilical vein endothelial cells (HUVECs) or primary murine aortic endothelial cells (mECs) were induced to undergo apoptosis through exposure to serum-free medium in vitro. In keeping with the translational scope of our work, we chose serum starvation for stimulation of apoptosis because exposure to serum-free preservation solution before transplantation is an integral component of all types of organ transplantations. In both cell types, serum deprivation significantly increased caspase-3 activation and the percentage of cells with chromatin condensation in the absence of cell membrane permeabilization (fig. S1, A and B). These results are consistent with previous work by our group demonstrating that this system induces apoptosis in the absence of primary or secondary necrosis (39, 42). Preventing activation of caspases and, more specifically, caspase-3 through biochemical inhibition or the use of mECs from caspase-3−/− mice significantly reduced the development of apoptosis (fig. S1, A and B). Medium conditioned by serum-starved HUVECs or mECs was harvested at time points associated with caspase-3 activation and evaluated for the presence of EVs (Fig. 1, A and B). Using a high-sensitivity flow cytometer equipped with an FSC coupled to a PMT designed to resolve smaller particles (fig. S2) (43, 44), we confirmed that vesicles of various sizes are released by ECs during apoptosis (Fig. 1, A and B). Most of the detected particles (95 to 99%) were sensitive to dissolution by the detergent Triton X-100, confirming that these structures are bound by a lipid membrane (Fig. 1, A and B, and fig. S3).

Fig. 1. Apoptotic ECs secrete two discrete types of EVs, apoptotic bodies and exosome-like vesicles, downstream of caspase-3 activation.

(A and B) Left panels: Representative side scatter (SSC)/forward scatter–photomultiplier tube (FSC-PMT) plots of the EVs detected with fluorochrome-conjugated annexin V in the supernatant of serum-starved HUVECs exposed to vehicle, DEVD, or ZVAD (A) or mECs from wild-type (WT) or caspase-3 knockout (Casp-3 KO) mice (B). Two major subpopulations, exosome-like vesicles (exo-like) and apoptotic bodies (apo bodies), are observed and gated. Triton panels represent annexin V+ events detected in the supernatant after treatment with 0.05% Triton X-100 (A) or mECs (B). Right panels: Flow cytometric quantifications of annexin V+ apoptotic bodies (white bars) and exosome-like vesicles (black bars) detected in the supernatant of HUVECs (A) or mECs (B). (C) Electron micrographs of apoptotic bodies (left) and exosome-like vesicles (right) released by serum-starved HUVECs and isolated by sequential ultracentrifugation. (D) Heatmap representation of protein abundance in isolated exosome-like vesicles and apoptotic bodies. Exosome-like vesicles and apoptotic bodies are characterized by distinct protein patterns. Perlecan is enriched in exosome-like vesicles, whereas several histones are unique to or enriched in apoptotic bodies. n ≥ 6 for each condition. Data are pooled from three independent experiments; mean ± SEM; t test.

As previously published, serum-starved ECs engage a mixed autophagic and apoptotic response without any necrosis (39, 42). Preventing autophagy through biochemical inhibition with bafilomycin A1 did not modulate the release of EVs (fig. S4). However, pan-caspase inhibition, specific caspase-3 inhibition, or the use of ECs from caspase-3−/− mice led to a significantly reduced production of apoptotic exosome-like vesicles and, to a lesser extent, apoptotic bodies (Fig. 1, A and B), highlighting the importance of caspases in the secretion of EVs by serum-starved ECs.

We then used sequential centrifugation of conditioned medium to isolate fractions enriched in apoptotic bodies or apoptotic exosome-like vesicles. The two fractions showed EVs with distinct ultrastructural characteristics (Fig. 1C and figs. S5 and S6). As reported previously (42), most of the EVs isolated by centrifugation at 50,000g were within the range of apoptotic bodies from 1 to 5 μm (containing intracytoplasmic components and various organelles, such as mitochondria) along with some smaller membrane vesicles within microvesicle size range (between 0.25 and 1 μM) (Fig. 1C and fig. S5). The fraction isolated by centrifugation at 200,000g was enriched in smaller EVs showing a heterogeneous population of membrane-bound vesicles ranging in size from 30 to 100 nm (Fig. 1C and fig. S6). Collectively, these results demonstrate that EVs of various sizes are released by apoptotic ECs through caspase-3–dependent pathways.

Proteomic profiling of exosome-like vesicles and apoptotic bodies reveals distinct proteome signatures

To further characterize the differences between EVs released by apoptotic ECs, we performed a large-scale proteomic analysis and profiled the abundance of proteins in apoptotic bodies and apoptotic exosome-like vesicles. Apoptotic bodies and exosome-like vesicles were characterized by strikingly distinct protein markers (Fig. 1D), confirming that exosome-like vesicles do not merely represent the debris of apoptotic bodies. Apoptotic bodies contain a higher number of proteins, with 546 identified proteins, whereas only 235 proteins were identified in exosome-like vesicles. The heatmap of protein abundance revealed that typical markers of apoptotic bodies such as histones (histone H2A type 2-A. histone H4, and histone H2B type 2-E) were enriched in apoptotic bodies (Fig. 1D).

We then performed gene ontology analysis to gain further insights into the processes by which proteins and pathways are actively modulated in the different EV subsets (fig. S7). In line with our previous findings (42), apoptotic bodies were found to be enriched in ribosomal proteins and proteins from the cytosol, endoplasmic reticulum, nucleus, and mitochondria. Exosome-like vesicles were enriched in extracellular matrix, basement membrane proteins, and several lysosomal proteins (fig. S7). More than 80% of the exosome-like vesicle proteome was listed in Exocarta and Vesiclepedia databases (fig. S8, A and B). Also, 62% of the top 50 exosome protein markers according to the Exocarta database were identified in exosome-like vesicles (table S1). Furthermore, Western blot analyses comparing the protein profiles of apoptotic bodies and apoptotic exosome-like vesicles showed enrichment in the exosome markers syntenin, fibronectin, and TCTP (fig. S8C). Apoptotic bodies did not express characteristic exosome markers but showed increased expression of GM130 and tubulin. However, some classical exosome markers such as CD9, CD81, and TSG101 were not expressed in apoptotic exosome-like vesicles, as evidenced by proteomic analysis or Western blotting, suggesting that these membrane vesicles, although closely related to exosomes, present some differences within protein signatures (fig. S8). The functions of proteins exclusive to exosome-like vesicles were associated with proteasomal degradation and ligase activities, whereas those of apoptotic bodies were enriched for RNA processing and vesicle targeting. More detailed analyses identified perlecan as one of the basement membrane proteins present in apoptotic exosome-like vesicles (Fig. 1D and fig. S7), with higher sequence coverage of the LG3 fragment of perlecan in exosome-like vesicles compared with apoptotic bodies (Fig. 2A and fig. S9). Because LG3 and antibodies to LG3 have been associated with rejection in animal models and transplant patients (8, 4547), we used electron microscopy with immunogold labeling to further characterize the presence of LG3 in preparations of apoptotic exosome-like vesicles. Immunogold labeling identified LG3 in vesicular structures ranging in size from 30 to 100 nm (Fig. 2B). To confirm that LG3 is released through caspase-dependent pathways, we assessed LG3 levels in preparations of exosome-like vesicles from caspase-inhibited or caspase3−/− ECs. Pan-caspase inhibition or the use of caspase-3−/− ECs significantly reduced the release of LG3 (Fig. 2, C and D). Collectively, these results demonstrate that apoptotic bodies and apoptotic exosome-like vesicles are characterized by distinct protein markers and that LG3 is a marker of apoptotic exosome-like vesicles.

Fig. 2. Apoptotic exosome-like vesicles are specifically enriched in perlecan-LG3 fragment and trigger the production of anti-LG3 autoantibodies.

(A) Sequence of the LG3 fragment and representation of identified peptides (red) in exosome-like vesicles and apoptotic bodies. The sequence coverage of LG3 is higher in exosome-like vesicles (59%) versus apoptotic bodies (28%). (B) Electron micrographs showing immunogold labeling of LG3 in exosome-like vesicles from serum-starved HUVECs. Data are pooled from three independent experiments. (C) Representative immunoblot and densitometric analysis of LG3 in exosome-like vesicles or apoptotic bodies purified from serum-starved apoptotic HUVECs incubated with the pan-caspase inhibitor Z-VAD or vehicle. (D) Representative immunoblot and densitometric analysis of LG3 in exosome-like vesicles or apoptotic bodies purified from serum-starved apoptotic mECs from WT or caspase-3 knockout mice. (E) Anti-LG3 IgG titers in sera from WT mice after 3 weeks of intravenous injections with vehicle, exosome-like vesicles, or apoptotic bodies purified from serum-starved mECs. n ≥ 6 for each condition; mean ± SEM; t test. OD, optical density.

Apoptotic exosome-like vesicles trigger anti-LG3 production

We then evaluated whether apoptotic exosome-like vesicles can trigger the production of anti-LG3 autoantibodies and shape the severity of rejection. We first evaluated whether intravenous injection of apoptotic exosome-like vesicles fosters the production of anti-LG3 antibodies in wild-type naïve mice. Sequential centrifugation was used to purify apoptotic bodies and apoptotic vesicles from serum-free medium conditioned by 1 × 104 murine apoptotic ECs. Each fraction was then resuspended in half of the initial volume of conditioned medium and intravenously injected into mice every other day for 3 weeks. Injection of apoptotic exosome-like vesicles, but not apoptotic bodies, significantly increased the production of anti-LG3 IgG antibodies in naïve mice (Fig. 2E). Because total protein levels were found to be six times lower in preparations of apoptotic bodies compared with preparations of apoptotic exosome-like vesicles (2 and 0.3 μg/ml, respectively), we also evaluated whether increasing the amount of injected apoptotic bodies would trigger the production of anti-LG3 antibodies. Injecting equal amounts of proteins from preparations of apoptotic bodies and preparations of apoptotic exosome-like vesicles failed to unmask immunogenic activity in apoptotic bodies (fig. S10).

We then evaluated whether injection of apoptotic exosome-like vesicles increases the production of other autoantibodies and markers of B cell activation. Anti-nuclear antibody (ANA) titers were increased in mice injected with exosome-like vesicles compared with mice injected with apoptotic bodies or control mice (fig. S11A). Enhanced numbers of germinal center B cells and follicular helper T cells were observed in mice injected with exosome-like vesicles (fig. S11, B to E). Collectively, these data suggest that apoptotic exosome-like vesicles, unlike apoptotic bodies, favor B cell responses associated, at least in part, with the production of autoantibodies.

Apoptotic exosome-like vesicles aggravate vascular rejection

We then compared the impact of apoptotic exosome-like vesicles and apoptotic bodies on the production of anti-LG3 antibodies and the severity of rejection in allografted mice. Apoptotic bodies and apoptotic exosome-like vesicles were purified from serum-free medium conditioned by 1 × 104 murine ECs and injected intravenously every other day for 3 weeks after transplantation into recipients of a fully major histocompatibility complex (MHC)–mismatched aortic graft. Because we aimed to evaluate aggravation of rejection, allografts were studied at 3 weeks after transplantation, a time point normally associated with minimal vascular remodeling and inflammation (fig. S12) (46, 47). The injection of transplanted mice with apoptotic exosome-like vesicles, but not apoptotic bodies, significantly increased the production of anti-LG3 IgG antibodies (Fig. 3A) and heightened the severity of rejection (Fig. 3, B to F).

Fig. 3. Apoptotic exosome-like vesicles aggravate vascular rejection.

(A) Anti-LG3 IgG titers in sera harvested 3 weeks after surgery from allografted mice injected with vehicle, exosome-like vesicles, or apoptotic bodies for 3 weeks after transplantation. (B) Hematoxylin and eosin (H&E), CD3, CD20, and C4d staining of aortic allograft sections from mice injected with vehicle, exosome-like vesicles, or apoptotic bodies for 3 weeks after transplantation (magnification: top panels, ×5; lower panels, ×20). (C) Intima/media ratio in allografts from mice injected with vehicle, exosome-like vesicles, or apoptotic bodies for 3 weeks after transplantation. (D) Mean number of CD3+ cells (T cells) per high-power field in allografts from mice injected with vehicle, exosome-like vesicles, or apoptotic bodies for 3 weeks after transplantation. (E) Mean number of CD20+ cells (B cells) per high-power field in allografts from mice injected with vehicle, exosome-like vesicles, or apoptotic bodies for 3 weeks after transplantation. (F) C4d deposition in allografts from mice injected with vehicle, exosome-like vesicles, or apoptotic bodies for 3 weeks after transplantation. (G) DSA levels in sera harvested 3 weeks after surgery from allografted mice injected with vehicle, exosome-like vesicles, or apoptotic bodies for 3 weeks after transplantation. n ≥ 8 for each condition. Data are pooled from three independent experiments; mean ± SEM; comparison with vehicle, t test.

Infiltration by CD3+ T cells and CD20+ B cells was significantly increased in mice injected with apoptotic exosome-like vesicles but not with apoptotic bodies (Fig. 3, B, D, and E). C4d deposition, a hallmark of immune complex deposition, was also increased in mice injected with apoptotic exosome-like vesicles (Fig. 3, B and F). DSAs were detected in allografted mice, but their levels were not affected by the injection of exosome-like vesicles or apoptotic bodies (Fig. 3G). This finding suggests that apoptotic exosome-like vesicles accelerate rejection through immune complex–dependent but DSA-independent pathways. The anti-LG3 antibodies produced secondary to injection of apoptotic exosome-like vesicles are mainly of complement-fixing isotypes (fig. S13). Collectively, these results demonstrate that apoptotic exosome-like vesicles, unlike apoptotic bodies, aggravate vascular allograft inflammation, C4d deposition, and production of complement-fixing isotypes of anti-LG3 antibodies.

The proteasome core complex is enriched and active in apoptotic exosome-like vesicles

The observation that exosome-like vesicles induce the production of anti-LG3 antibodies in both grafted and naïve mice prompted us to further analyze their protein signature. Intriguingly, we observed a strong enrichment for the proteasome core complex specifically in apoptotic exosome-like vesicles (Fig. 4, A and B). In our proteomic data set, all the catalytic subunits of the 20S proteasome core were identified with high confidence in the exosome-like vesicles, whereas these proteins were absent in apoptotic bodies (Fig. 4, A and B). Western blot analysis further confirmed the presence of the α3 subunit of the proteasome specifically in exosome-like vesicles (Fig. 4A). In contrast, only four proteasome-associated proteins were identified in apoptotic bodies, all contained within the 19S regulatory subunit. Immunogold labeling of the 20S proteasome identified immunoreactivity in EVs ranging in size from 30 to 100 nm (Fig. 4C).

Fig. 4. The proteasome, active within apoptotic exosome-like vesicles, regulates the processing of various proteins, including LG3.

(A) Heatmap representation of proteasome protein abundance in exosome-like vesicles and apoptotic bodies. Bottom: Western blot analysis of the α3 proteasome subunit in exosome-like vesicles (exo-like) and apoptotic bodies (apo bodies). (B) Schematic representation of the proteins detected specifically in exosome-like vesicles (red) and in apoptotic bodies (black). The complete proteasome 20S core complex is found in exosome-like vesicles, whereas only regulatory subunits are detected in apoptotic bodies. (C) Electron micrographs showing immunogold labeling of the α3 proteasome subunit in exosome-like vesicles from serum-starved HUVECs. (D) Quantification of proteasome chymotrypsin, trypsin, and caspase-like proteolytic activities in serum-starved HUVECs (gray bars), exosome-like vesicles (black bars), and apoptotic bodies (white bars). n ≥ 5 for each condition; mean ± SEM; comparison with cells. (E) Quantification of proteasome chymotrypsin, trypsin, and caspase-like proteolytic activities in exosome-like vesicles from serum-starved HUVECs treated with the proteasome inhibitor bortezomib or vehicle. Means ± SEM; comparison with vehicle. (F) Flow cytometric quantification of annexin V+ EVs secreted by serum-starved HUVECs treated with the proteasome inhibitor bortezomib or vehicle. (G) Heatmap representation of protein abundances in exosome-like vesicles from serum-starved HUVECs treated with the proteasome inhibitor bortezomib or vehicle. Bortezomib treatment strongly modulates the protein content of exosome-like vesicles. Perlecan is enriched in bortezomib-treated cells. (H) LG3 spectral count in exosome-like vesicles from serum-starved HUVECs treated with the proteasome inhibitor bortezomib or vehicle. (I) Proteasome inhibition with bortezomib results in the accumulation of ubiquitylated proteins in exosome-like vesicles. Upon bortezomib treatment, 20 ubiquitylation sites on 19 proteins were detected, whereas in vehicle-treated control cells, only five ubiquitinylation sites on five proteins were detected by MS. Mean ± SEM; t test; ns, nonsignificant.

Quantitative proteomics based on spectral count indicated comparable levels of all three catalytic subunits containing chymotrypsin-like (β5), trypsin-like (β2), and caspase-like (β1) activities in exosome-like vesicles (table S1). However, we detected several trypsin inhibitors (serpins) in our data set (table S1). Notably, ITIH2 (inter-α-trypsin inhibitor heavy chain H2) is exclusively found in exosome-like vesicles. SERPINE1 (plasminogen activator inhibitor 1) is also enriched in exosome-like vesicles, whereas it is essentially absent in apoptotic bodies. The presence of inhibitors of specific enzymatic activities in the exosome-like vesicles could result in a dynamic modulation of the proteasome activity. Thus, we evaluated whether proteasome activity was present in the various fractions. Consistent with our proteomic findings, we found enhanced caspase-like activity per microgram of total proteins in the apoptotic exosome-like vesicle fraction compared to apoptotic bodies and parent cells (Fig. 4D). The enhancement of caspase-like activity over other proteolytic functions of the proteasome correlated with the identification of trypsin and chymotrypsin inhibitors in the apoptotic nanovesicle fraction.

To evaluate whether the presence of active proteasome is a major characteristic of apoptotic exosome-like vesicle, we evaluated if healthy ECs could also secrete membrane vesicles containing active 20S proteasome. To avoid contamination of culture medium by membrane vesicles derived from fetal bovine serum (FBS), we generated vesicle-free normal medium using FBS spun at 200,000g before its addition to culture medium (fig. S14A). Endothelial cells were exposed to normal culture medium for 4 hours. Small-particle flow cytometry demonstrated the presence of exosome-size membrane vesicles in medium conditioned by healthy cells (fig. S14B). In comparing by Western blot equal amounts of membrane vesicles produced by normal or serum-starved apoptotic ECs and isolated by centrifugation at 200,000g, we found that the exosome markers fibronectin, syntenin, and TCTP were present in both conditions (fig. S14D). However, the 20S proteasome core complex and caspase-like proteasome activity were detected exclusively in membrane vesicles released by serum-starved apoptotic ECs (fig. S14, C and D). These results support the notions that active 20S proteasome is secreted within exosome-like vesicles specifically under proapoptotic serum-starved conditions.

We then investigated if the presence of an active proteasomal core complex in apoptotic exosome-like vesicles is specific to apoptotic EC–derived vesicles. To this end, we evaluated preparations of apoptotic exosome-like vesicles from medium conditioned by serum-starved vascular smooth muscle cells and tubular epithelial cells undergoing similar levels of apoptosis as serum-starved ECs (fig. S15, A and B). Preparations from both cell types highlighted the presence of the 20S proteasome core complex by Western blot analysis as well as proteasome activity (fig. S15, C and D), demonstrating that the presence of an active proteasome core complex is not exclusive to endothelial apoptotic exosome-like vesicles.

To investigate whether the activity of the proteasome modulates the release and immunogenic activity of apoptotic exosome-like vesicles, we exposed ECs to bortezomib, a proteasome inhibitor, during serum starvation. Proteasome inhibition neither modulated endothelial apoptosis (fig. S16) nor prevented the production of EVs but markedly reduced caspase-like activity in exosome-like vesicles (Fig. 4, E and F). We also performed large-scale proteomics on purified apoptotic exosome-like vesicles from bortezomib- or vehicle-treated apoptotic ECs. Bortezomib treatment strongly modulated the protein content of exosome-like vesicles and increased the number of proteins identified (Fig. 4G). Proteasome inhibition resulted in the enrichment of LG3 within exosome-like vesicles, suggesting that LG3 is a proteasome substrate (Fig. 4H). Whereas 154 proteins were found in control exosome-like vesicles, 337 proteins were identified in bortezomib-treated exosome-like vesicles (table S2). Also, 198 proteins were detected exclusively in exosome-like vesicles from bortezomib-treated cells, suggesting an important role for the proteasome in the degradation of vesicle proteins (table S2). Proteasomal degradation occurs in a ubiquitin-dependent manner. We therefore identified ubiquitylated proteins by mass spectrometry (MS). Proteasome inhibition with bortezomib resulted in the accumulation of ubiquitylated proteins in exosome-like vesicles (Fig. 4I). Upon bortezomib treatment, 20 ubiquitylation sites on 19 proteins were detected in nanovesicle extracts, whereas exosome-like vesicles produced in vehicle-treated serum-starved cells showed only 5 ubiquitylation sites on 5 proteins (table S3). Collectively, these results support the notion that the proteasome, active within apoptotic exosome-like vesicles, regulates the processing of various proteins, including LG3.

Proteasome activity is required for induction of anti-LG3 production and aggravation of allograft inflammation

Because LG3 behaves as a proteasome substrate, we investigated the possibility that the proteasome activity of apoptotic exosome-like vesicles regulates their capacity to induce the production of anti-LG3 autoantibodies. Both allografted and nongrafted mice were injected with an equal number of apoptotic exosome-like vesicles purified from either bortezomib-treated serum-starved ECs or control vehicle-treated serum-starved ECs. Naïve and transplanted mice exposed to apoptotic exosome-like vesicles from bortezomib-treated cells showed reduced levels of anti-LG3 antibodies when compared to control exosome-like vesicles (Fig. 5A). In transplanted mice, neointimal remodeling was similar in mice injected with exosome-like vesicles from bortezomib-treated cells or control apoptotic cells (Fig. 5B). However, allograft infiltration by CD3+ T cells and CD20+ B cells and C4d deposition were significantly decreased in mice injected with apoptotic exosome-like vesicles from bortezomib-treated cells (Fig. 5, B to D). Collectively, these results suggest that proteasome activity within apoptotic exosome-like vesicles regulates anti-LG3 autoantibody formation and allograft inflammation.

Fig. 5. Proteasome activity within apoptotic exosome-like vesicles regulates anti-LG3 autoantibody formation and allograft inflammation.

(A) Anti-LG3 IgG titers in sera from WT (gray bars) or allografted mice (black bars) after 3 weeks of intravenous injections with vehicle or with exosome-like vesicles from serum-starved mECs treated with the proteasome inhibitor bortezomib (exo-like bortezo) or vehicle (exo-like vehicle). (B) CD3-, CD20-, and C4d-stained aortic allograft sections from mice injected with exosome-like vesicles from serum-starved mECs treated with bortezomib or vehicle (magnification, ×20). (C) Mean number of CD3+ cells (T cells) per high-power field in allografts from mice injected with exosome-like vesicles from serum-starved mECs treated with bortezomib or vehicle for 3 weeks after transplantation. (D) Mean number of CD20+ cells (B cells) per high-power field in allografts from mice injected with exosome-like vesicles from serum-starved mECs treated with bortezomib or vehicle for 3 weeks after transplantation. (E) C4d deposition in allografts from mice injected with exosome-like vesicles from serum-starved mECs treated with the proteasome inhibitor bortezomib or vehicle for 3 weeks after transplantation. n ≥ 9 for each condition. Data are pooled from four independent experiments; mean ± SEM; if not indicated, comparison with vehicle, t test.

Vascular injury increases proteasome-activity in circulating exosome-like vesicles and triggers anti-LG3 production in vivo

We then evaluated whether this pathway is recapitulated in vivo after vascular injury. We first focused on renal artery clamping, a model of acute kidney injury (AKI), associated with increased caspase-3 activation within the renal microvasculature (4851). We chose this model because anti-LG3 antibodies have been found in renal disease patients awaiting a kidney transplant, and AKI is common in these patients and predicts progressive renal dysfunction (52). In this model, renal function was reduced up to 7 days after ischemia-reperfusion injury (Fig. 6, A and B). The 20S proteasome core complex was identified within circulating exosome-like vesicles purified from mouse serum 2 days after AKI by immunogold labeling but was absent from exosome-size circulating vesicles purified from normal mice serum (Fig. 6G and fig. S17). Moreover, both proteasome caspase-like activity and LG3 levels increased in exosome-like vesicle preparations after AKI (Fig. 6, C, E, and F). Anti-LG3 IgG antibody titers increased significantly 7 and 14 days after surgery and returned to baseline at 21 days (Fig. 6D).

Fig. 6. Renal injury enhances circulating proteasome caspase-like activity, LG3 levels, and anti-LG3 production.

(A) Blood urea nitrogen (BUN) concentration in mice pre-surgery or at 2, 7, or 21 days after renal ischemia-reperfusion (I/R) injury. (B) Serum creatinine concentration in mice pre-surgery or at 2, 7, or 21 days after renal ischemia-reperfusion injury. (C) Quantification of proteasome caspase-like activity in exosome-like vesicles from serum pre-surgery or 2 days after renal ischemia-reperfusion injury. (D) Serum anti-LG3 IgG titers pre-surgery or at 2, 7, 14, and 21 days after renal ischemia-reperfusion injury. (E) Densitometric analyses of immunoblots for LG3 detection performed on exosome-like vesicles purified from sera of mice pre-surgery or 2 days after renal ischemia-reperfusion injury. (F) Representative Ponceau Red staining and immunoblot showing the detection of perlecan fragments, including LG3, in serum exosome-like vesicles pre-surgery or 2 days after renal ischemia-reperfusion injury. (G) Left panel: Electron micrographs of exosome-like vesicles isolated from serum 2 days after renal ischemia-reperfusion injury. Right panels: Electron micrographs showing immunogold labeling of the α3 proteasome subunit in exosome-like vesicles isolated from serum 2 days after renal ischemia-reperfusion injury. n = 6 for each condition; mean ± SEM; comparison with pre-surgery, t test.

We studied a second model of vascular injury evoked by femoral arteriectomy leading to hindlimb ischemia (Fig. 7A), taking into consideration the prevalence of severe peripheral vascular disease in the end-stage renal disease population. Again, proteasome caspase-like activity and LG3 levels in circulating exosome-like vesicles increased significantly after femoral arteriectomy (Fig. 7, B to D). Anti-LG3 IgG titers also increased significantly 7 days after surgery and remained elevated for up to 21 days (Fig. 7E). Collectively, these results demonstrate that vascular injury in vivo leads to enhanced caspase-like proteasome activity in circulating exosome-like vesicles followed by the production of anti-LG3 antibodies.

Fig. 7. Vascular injury enhances circulating proteasome caspase-like activity, LG3 levels, and anti-LG3 production.

(A) Laser Doppler imaging of blood flow in the lower limbs of mice on the day of the surgery (day 0) or at 7 or 21 days after femoral arteriectomy (vascular injury). (B) Quantification of proteasome caspase-like activity in serum exosome-like vesicles pre-surgery or 2 days after vascular injury. (C) Representative Ponceau Red staining and immunoblot showing the detection of perlecan fragments, including LG3, in serum exosome-like vesicles pre-surgery or 2 days after vascular injury. (D) Densitometric analyses of immunoblots for LG3 detection performed on serum exosome-like vesicles pre-surgery or 2 days after vascular injury. (E) Serum anti-LG3 IgG titers pre-surgery or at 7, 14, and 21 days after vascular injury. n = 6 for each condition; mean ± SEM; comparison with pre-surgery, t test.

DISCUSSION

Apoptosis is classically considered an anti-inflammatory and tolerogenic type of cell death (34). However, mounting evidence shows that the presence of autoantibodies reactive with components of apoptotic cells before solid organ transplantation predicts reduced allograft survival and risk of rejection (8, 13, 18, 19, 30). The present study identifies apoptotic exosome-like vesicles as a novel and immunogenic component of the paracrine apoptotic response. Using electron microscopy, unbiased proteomics, and functional studies, we show that nanoscale EVs released by apoptotic cells display immune-mediating functions distinct from apoptotic bodies. Apoptotic exosome-like vesicles induce B cell responses in naïve and transplanted mice leading to the production of autoantibodies. Apoptotic exosome-like vesicles, but not classic apoptotic bodies, aggravate vascular rejection, with increased allograft infiltration by CD3+ T cells and CD20+ B cells, C4d deposition, and vascular remodeling. C4d deposition induced by exosome-like vesicles is likely dependent on non-HLA antibodies because exosome-like vesicles did not modulate the levels of anti-HLA antibodies in rejecting mice and triggered the production of complement-fixing autoantibodies such as anti-LG3, known to promote C4d deposition (8).

Our results highlight the complexity of paracrine signals released by apoptotic cells downstream of caspase-3 activation with the concomitant release of tolerogenic and immunogenic signals. Unbiased comparative proteomic analysis established that different and almost mutually exclusive sets of proteins are present in apoptotic bodies and apoptotic exosome-like vesicles, suggesting that their generation requires differential sorting of proteins and that their biogenesis is likely different. Whereas it is well documented that caspase-3 activation regulates the generation of apoptotic bodies by plasma membrane blebbing (53), our results propose an important role for caspase-3 activation in the release of apoptotic exosome-like vesicles. Although the precise molecular components regulating differential sorting of proteins into either apoptotic bodies or exosome-like vesicles are unclear, the present results suggest that distinct molecular pathways activated downstream of activated caspase-3 lead to the production of either apoptotic bodies or apoptotic exosome-like vesicles. Whether nanovesicles are shed from the plasma membrane, generated by exocytosis of multivesicular bodies (MVBs) in a caspase-dependent manner or by another mechanism, remains to be determined. We showed in previous work that serum starvation leads to the accumulation of multivesicular structures (39), but the link between the biogenesis of apoptotic exosome-like nanovesicles and MVBs requires further investigations. Apoptotic exosome-like vesicles were found to express a protein signature that is closely related yet distinct from classical exosomes. The presence and absence of different sets of proteins have been suggested as minimal requirements to characterize EVs (54). Transmembrane- or lipid-bound extracellular proteins (such as CD9, CD63, CD81, integrins, cell adhesion molecules, growth factor receptors, heterotrimeric guanine nucleotide–binding proteins, and MFG-E8) and cytosolic proteins (TSG-101, annexins, Rabs, and syntenin) and the underrepresentation of endoplasmic reticulum, Golgi, and mitochondrial and nuclear proteins are considered characteristic of exosomes (54). Our proteomic data identified integrins [integrin α5 (ITGA5) and ITGB1], adhesion molecules [melanoma cell adhesion molecule (MCAM)], annexins [annexin A1 (ANXA1), ANXA2, and ANX5], and syntenin in our preparations and confirmed the underrepresentation of endoplasmic reticulum, Golgi, and mitochondrial and nuclear proteins, suggesting that small vesicles released by apoptotic cells display typical characteristics of exosomes. Our results also highlighted some distinctive features such as release through caspase-3–dependent pathways and the presence of an active proteasome core complex in small vesicles released by apoptotic cells, hence our choice to refer to them as “apoptotic exosome-like” vesicles and not as classical exosomes.

One of the most striking observations of this study is the identification of proteasome activity in endothelial exosome-like vesicles. In eukaryotic cells, proteasomes are highly conserved protease complexes typically composed of the 19S regulatory cap and the 20S proteolytic core. The proteasomal degradation pathway is essential for many cellular processes, including the cell cycle, the regulation of gene expression, and antigenic presentation. In humans, extracellular proteasomes have been found circulating in the plasma of patients suffering from a variety of inflammatory, autoimmune, and neoplastic diseases (5560), and in various conditions, the concentration of circulating proteasomes correlates with disease activity (61). The presence of circulating proteasome is intriguing because proteasomal subunits do not contain signal sequence for export via the classic secretory pathway, which argues for an alternative secretion mechanism. The release of proteasome in the extracellular milieu as a result of membrane disruption is proposed and cannot be excluded, but the present results and other studies support the existence of regulated mechanisms. The release of active 20S proteasome has also been described recently in EVs originating from other cell types such as T cells (shedding microparticles) (62) and mesenchymal stem cells (in exosomes derived from the endosomal compartment) (63). Here, we demonstrate that the release of active proteasome within exosome-like vesicles occurs in association with cell stress and stimulation of apoptosis because healthy ECs do not secrete active proteasome within membrane vesicles. However, the secretion of the proteasome core complex in apoptotic vesicles is not specific to ECs because we found the presence and activity of the proteasome in nanovesicles produced by other cell types undergoing apoptosis such as vascular smooth muscle cells and tubular epithelial cells, suggesting that this pathway is likely applicable to various disease states associated with apoptotic cell death.

Our results also show that proteasome activity within apoptotic exosome-like vesicles is one of the key determinants of their immunogenicity. Inhibiting proteasome activity in parent apoptotic ECs did not modulate cell death and the release of apoptotic exosome-like vesicles, but completely abrogated all proteasome activity in exosome-like vesicles. Injection of exosome-like vesicles with inhibited proteasome activity elicited significantly less anti-perlecan/LG3 production when injected in naïve mice. Injection of these vesicles in transplanted animals also resulted in reduced levels of anti-LG3 antibodies as well as lower numbers of graft-infiltrating T and B cells and reduced C4d deposition. Reduced anti-perlecan/LG3 production occurred in spite of higher levels of perlecan/LG3 in proteasome-inhibited vesicles, demonstrating that failure to trigger an antibody-response is not caused by lower levels of the immunogen. Although detailed characterization of the specific molecular components regulating perlecan/LG3 immunogenicity by the active proteasome core within exosome-like vesicles will require further investigation, our results demonstrate that proteasome activity within these vesicles is central to the induction of anti-perlecan/LG3 production in vivo. Bortezomib is currently being used in the treatment of refractory ABMR as a B cell–depleting agent based on its proapoptotic activity on plasma cells and B cells (64, 65). The present results add further insights into additional pathways of potential importance in the anti-rejection activity of bortezomib. By lowering the immunogenicity of exosome-like vesicles produced at sites of vascular injury, bortezomib could prevent the formation of autoantibodies that further accelerate leukocyte infiltration and complement deposition. However, the proteasome activity of exosome-like vesicles was dispensable for acceleration of neointima formation. Various factors could explain this finding. The perlecan fragment LG3 per se acts as a promigration factor in mesenchymal stem cells and smooth muscle cells, thus promoting their accumulation in sites of vascular injury (46, 47), and perlecan fragments have also been reported to bind Toll-like receptor 2 (TLR2) and TLR4 (66). Other proteins and mediators present in exosome-like vesicles, such as tissue-degrading enzymes (metalloproteinases 2 and 9), DAMPs/TLR ligands (heat shock proteins 75, 10, 71, and 70 and fibronectin), and PAI-1 (SERPINE1), could also potentially accelerate the fibroproliferative and inflammatory responses of the vessel wall. Moreover, studying the nucleic acid and lipid content of these exosome-like vesicles might offer additional insights into the mechanisms induced by apoptotic exosome-like vesicles.

Humoral immune responses against perlecan have been found to be associated with rejection in several human transplant cohorts and in animal models of acute and chronic allograft rejection (8, 21, 67, 68). In all these settings, ischemia-reperfusion injury has been suspected to play a role in anti-perlecan autoimmune responses. Because anti-perlecan antibodies have been described before transplantation in patients without autoimmune conditions (29), we evaluated whether episodes of vascular injury, which are common in renal disease patients, could be associated with increased circulating levels of markers characteristic of apoptotic exosome-like vesicles. Two models of vascular injury in mice led to increased caspase-like activity and LG3 levels in circulating exosome-like vesicles followed by elevated anti-LG3 IgG levels, confirming a close association between vascular and/or tissue injury, release of immunogenic exosome-like vesicles, and production of anti-LG3.

Our study, however, has a number of limitations. First, the biogenesis of apoptotic exosome-like vesicles remains incompletely understood. The abundance of exosomal markers suggests that inward budding of MVBs could be contributing to their biogenesis, yet how caspase-3 affects intracellular trafficking and membrane fusion events leading to their release will require further investigation. Second, although we have shown that circulating levels of proteasome-labeled vesicles increase after renal ischemia-reperfusion injury and that proteasome caspase-like activity is increased in these vesicle preparations, we cannot precisely determine the cellular origin of these vesicles. Endothelial cells, vascular smooth muscle cells, and tubular epithelial cells all release proteasome-active vesicles when injured in vitro. These cell types are targets of injury during ischemia-reperfusion and could all be contributing to the release of apoptotic exosome-like vesicles in vivo. Further studies will be required to identify cell type–specific markers of apoptotic exosome-like vesicles. Finally, we have demonstrated that exosome-like vesicles accelerate vascular rejection without increasing the production of donor-specific alloantibodies but rather by triggering the formation of humoral autoimmune pathways. The current study focused primarily on the protein composition of exosome-like vesicles and apoptotic bodies and enabled the identification of source proteins from which could be derived antigenic peptides. It is also possible that these vesicles may contain small peptides from proteasome degradation that remained elusive to the current proteomic approach. The identification of these peptides and of specific perlecan-derived MHC class I or class II epitopes will require alternate methods that facilitate their recovery and further characterization by MS/MS sequencing to account for the diversity of the corresponding degradation products. This will be the subject of future investigations.

Collectively, these results support the notion that vascular injury in association with renal injury before transplantation leads to increased circulating proteasome activity within exosome-like vesicles fractions, therefore increasing sensitization to autoantigens. By providing novel insights into pathways responsible for the production of autoantibodies before transplantation, these results suggest that assessment of proteasome activity within circulating apoptotic exosome-like vesicles before and at the time of transplantation represents new avenues for predicting and controlling maladaptive humoral responses to apoptotic cell components that enhance the risk of rejection after transplantation.

MATERIALS AND METHODS

Study design

The aim of this study was to evaluate the immunogenic potential of the various extracellular structures released by apoptotic ECs. We use large-scale proteomics to profile the protein signatures of apoptotic bodies and apoptotic exosome-like vesicles generated in vitro. We then compared the capacity of apoptotic exosome-like vesicles and apoptotic bodies to trigger the production of autoantibodies in transplanted and nontransplanted mice and shape the severity of rejection. The observation that exosome-like vesicles induced the production of anti-LG3 antibodies in both grafted and naïve mice prompted us to further analyze their protein signature. We found a strong enrichment and activity of the 20S proteasome core complex in apoptotic exosome-like vesicles by proteomic analyses, immunoblots, and electron microscopy. We next investigated, using inhibition of proteasome activity with bortezomib, whether the activity of the proteasome modulates the release and immunogenic activity of apoptotic exosome-like vesicles in vitro and in vivo. Finally, using in vivo models of vascular injury, we evaluated whether the release of proteasome-active exosome-like vesicles occurs in vivo after vascular injury. Group sizes were selected on the basis of our experience with these systems. Mice were age- and sex-matched between groups. Investigators were not blinded when conducting and evaluating the experiments, and no randomization was necessary. No data were excluded from this study.

Cell culture and conditioned medium preparation

Endothelial cells. mECs were isolated from the aorta of C57Bl/6 mice and cultured in low-glucose Dulbecco’s modified Eagle’s medium (DMEM) supplemented with EC growth supplements (Alfa Aesar), 10% FBS (Invitrogen), 10% new born calf serum (Invitrogen), heparin (6300 U; Sandoz), 1% penicillin-streptomycin, and 1% fungizone. HUVECs were obtained from Clonetics and cultured in EGM-2MV complete medium (Clonetics). To generate conditioned medium, cells were exposed to serum-free medium alone or in the presence of the pan-caspase inhibitor Z-VAD-fmk (100 μM; R&D Systems), the caspase-3 and caspase-7 inhibitor DEVD (50 μM; R&D Systems), the proteasome inhibitor bortezomib (100 μg/ml; Stressmarq), or vehicle (dimethyl sulfoxide) for 4 hours (HUVECs) or 9 hours (mECs). In previous work, we demonstrated that this system leads to the release of active mediators by apoptotic ECs downstream of caspase-3 activation without cell membrane permeabilization (40).

Human proximal tubular epithelial cells. Human proximal tubular epithelial cells (HPTC) were isolated as previously described from nondiseased nephrectomy samples from patients with renal cell carcinoma (69). Briefly, cortical tissue from kidney samples were incubated with collagenase (Sigma) for 1 hour and passed through a 70-μm cell strainer (Falcon). Cells were washed and resuspended in HPTC medium [10% FBS (Sigma), 1% penicillin-streptomycin (Sigma), 1% hormone mix (Sigma), human epidermal growth factor (25 ng/ml; Sigma), 25 mM Hepes (Sigma)], plated on collagen IV–coated plates, and grown at 37°C in 5% CO2. HPTCs were washed and the medium was changed the following day. To generate conditioned medium, cells were exposed to serum-free medium for 24 hours.

Vascular smooth muscle cells. A7R5 rat vascular smooth muscle cells (American Type Culture Collection) were cultured in DMEM/Nutrient Mixture F-12 supplemented with 10% FBS (Invitrogen). To generate conditioned medium, cells were exposed to serum-free medium for 48 hours.

Screening for apoptosis by fluorescence microscopy

Fluorescence microscopy of unfixed/unpermeabilized adherent cells stained with Hoechst 33342 (20-(4-ethoxyphenyl)-5-(4-methyl-1-piperazinyl)-2.50-bi-1-1H-benzimidazole) and propidium iodide was performed as previously described (42).

Immunoblotting

Proteins were extracted, separated by electrophoresis, transferred onto nitrocellulose membranes or polyvinylidene difluoride (for caspase-3 detection), and probed. The antibodies used for Western blotting were antibodies against perlecan (Santa Cruz Biotechnology), 20S proteasome α3 (Santa Cruz Biotechnology), CD63 (Abcam), GM130 (Abcam), tubulin (Calbiochem), syntenin-1 (Santa Cruz Biotechnology), TCTP (Santa Cruz Biotechnology), and fibronectin (Abcam).

Flow cytometric analyses of EVs

The flow cytometric analyses were done as previously described by Rousseau et al. (70). In brief, fluorescence was used as a trigger signal, and positive fluorescent events were plotted on an SSC/FSC-PMT graph. The nanovesicle gate of detection was determined on the basis of the acquisition of sky blue and yellow-green microspheres of sizes 90, 450, 840, 1000, and 3200 nm (fig. S1), and 1000 microspheres were acquired. Conditioned media were labeled for 30 min with V450 probe–conjugated annexin V (diluted 1:50; BD Biosciences) at room temperature in the dark. Because annexin V recognizes PS in a calcium-dependent manner, we confirmed its specific recognition of EVs by chelating Ca2+ ions using sample detection in the presence of 50 μM EDTA as a baseline (70). To process the data quantitatively, a known number of polystyrene microspheres (15-μm diameter; Polysciences) were added to each tube as previously described (70). To confirm that the events detected were genuine EVs and not protein aggregates, their sensitivity to 0.05% Triton X-100 was assessed.

Preparation of fractions enriched in apoptotic bodies or apoptotic exosome-like vesicles

Serum-free media conditioned by HUVECs were fractionated using sequential centrifugation: a first centrifugation at 1200g for 15 min at 4°C to pellet cell debris; a second centrifugation at 50,000g for 15 min at 4°C to pellet apoptotic bodies; and a final ultracentrifugation at 200,000g for 18 hours at 4°C to pellet exosome-like vesicles. Pellets containing either apoptotic bodies or exosome-like vesicles were resuspended in half of the initial volume of conditioned medium. For chronic exposure to the different apoptotic vesicles, a volume of 150 μl of either exosome-like vesicle or apoptotic body preparation or vehicle [phosphate-buffered saline (PBS)] was injected intravenously (caudal vein) every second day up to eight injections in the different animal models used (see “Animal studies”).

Proteasome activity assay

The proteasome activity assay was performed in white 96-well plates using the Proteasome-Glo Chymotrypsin-like, Trypsin-Like, and Caspase-Like Cell-Based Assays (Promega) according to the manufacturer’s instructions using cell lysate (0.4 μg), exosome-like vesicles (0.4 μg), or apoptotic bodies from HUVEC supernatant (0.4 μg) or exosome-like vesicles purified from mouse serum (50 μg). Luminescence was measured using a PerkinElmer Victor 3 V 1420 Multilabel Counter 1420-040 Microplate Reader.

Electron microscopy

Apoptotic bodies. The fraction enriched in apoptotic bodies was fixed in 2% glutaraldehyde–0.1 M sodium cacodylate, postfixed in 1% OsO4, dehydrated in alcohol, processed for flat embedding in Epon 812, and observed under a Zeiss CEM 902 electron microscope, as previously described (71).

Exosome-like vesicles. Preparations of exosome-like vesicles were fixed in 4% paraformaldehyde in PBS buffer and were either contrasted and embedded for whole-mount electron microscopic observation or processed for immunolabeling. Fixed exosome-like vesicles (5 μl) were deposited on a Formvar/carbon-coated grid for 20 min and washed on drops of PBS at room temperature. The grid was then directly processed for negative staining or immunolabeling. Negative staining: The grid was transferred onto a drop of 1% glutaraldehyde for 5 min and washed on several drops of water. Contrasting was performed by transferring the grid onto a drop of 2% uranyl acetate–0.075 M oxalate (pH 7) solution for 5 min. Finally, the grid was transferred onto a drop of 2% methyl cellulose–4% uranyl acetate (9:1) solution for 10 min on ice, and excess solution was blotted with Whatman filter paper and air-dried. Examination was performed with a Philips CM100 electron microscope. Immunolabeling: The grid was blocked by incubating in 50 mM glycine in PBS for 15 min and then transferred onto 1% ovalbumin in PBS for 10 min. The grid was incubated for 30 min on a drop of rabbit anti-LG3 antibody or rabbit anti-proteasome antibody, both diluted at 1:50 in PBS (Santa Cruz Biotechnology Inc.), and washed on several drops of PBS. Finally, the grid was incubated on a drop of anti-rabbit IgG conjugated to 10-nm gold particles for 20 min, washed with PBS, and processed for negative staining.

Proteomic analysis of purified exosome-like vesicles

SDS–polyacrylamide gel electrophoresis and MS analysis. Proteins (20 μg) were separated on a 4 to 12% precast NuPAGE gel (Invitrogen). The gel was Coomassie-stained, and the lanes were cut into pieces, reduced with dithiothreitol (Sigma-Aldrich), and alkylated by chloroacetamide (Sigma-Aldrich). The gel pieces were then digested by trypsin, and peptides were extracted three times with 90% acetonitrile (ACN)/0.5 M urea. Combined extracts were desalted with ZipTips (Millipore), dried, and resuspended in 5% ACN, 0.2% formic acid for MS analyses. Peptides were separated on a 150-μm inside diameter, 15-cm reversed-phase nano-LC column (Jupiter C18, 3 μm, 300 Å, Phenomex) with a loading buffer of 0.2% formic acid. Peptide elution was achieved by a gradient of 5 to 40% ACN in 85 min on an Eksigent 2D nano-LC operating at a flow rate of 600 nl/min. The nano-LC was coupled to an LTQ-Orbitrap Elite mass spectrometer (Thermo-Electron), and samples were injected in an interleaved manner. The mass spectrometer was operated in a data-dependent acquisition mode with a 1-s survey scan at 120,000 resolution, followed by 12 product ion scans (MS/MS) of the most abundant precursors above a threshold of 10,000 counts in the LTQ part of the instrument. CID was performed in the LTQ at 35% collision energy and an activation Q of 0.25. For each MS-based proteomics experiment, three independent biological replicates were used in triplicates.

Protein identification and data analysis. The centroided MS/MS data were merged into single peak-list files (Distiller, v2.4.2.0) and searched with the Mascot search engine v2.3.01 (Matrix Science) against the forward and reversed IPI human v3.54 database. Mascot was searched with a parent ion tolerance of 0.05 dalton and a fragment ion mass tolerance of 0.5 dalton. Carbamidomethylation of cysteine, oxidation of methionione, and deamidation and phosphorylation of serine, threonine, and tyrosine residues (and ubiquitylation of lysine residues (GlyGly) were specified as variable modifications. Proteins were considered identified when they had at least two different peptide identifications and the combined score of unique peptide identifications exceeded the score of the first reversed database hit reaching 2%. This resulted in a false discovery rate of <2% on the peptide level. Relative protein abundance was determined using a redundant peptide counting approach (spectral counts). The value of redundant peptide counts from three identification cycles (on three independent biological replicates) was used to generate heatmaps.

Bioinformatics. Gene Ontology (GO) annotations for cellular component, biological process, and molecular function were obtained from the GO project using the DAVID Bioinformatics resources (http://david.abcc.ncifcrf.gov/) (72, 73). To identify the GO terms that were statistically overrepresented in our protein list, we used the binomial statistics tool to compare the classifications of multiple clusters of lists to a reference list (Homo sapiens total proteome). Only the terms that were significantly enriched/depleted with a P value <0.05 were used for the analysis.

Animal studies

Adult C57Bl/6 and BALB/c mice (20 to 22 g; Charles River) were maintained on a 12-hour light-dark cycle and fed a normal diet ad libitum. All experiments on mice were approved by the Centre hospitalier de l’Université de Montréal (CRCHUM) Comité Institutionnel de Protection des Animaux (CIPA).

Aorta transplantation procedures

Mice were anesthetized using isoflurane (2%) by inhalation. Aortic transplantation was performed as described elsewhere with minor modifications (74, 75). Briefly, 1 ml of heparinized saline (50 μl/ml) was injected into the vena cava to flush the aorta. A 6-mm segment of abdominal aorta from below the renal arteries to just above the aortic bifurcation was excised and soaked in ice-cold 0.9% normal saline. When mentioned, warm ischemia was induced by clamping the aorta for 15 min before excision from the donor. The grafts were then excised and sutured orthotopically with end-to-end anastomoses using 11-0 nylon interrupted sutures.

Renal ischemia-reperfusion injury procedures

Mice were anesthetized with isoflurane and buprenorphin (0.05 mg/kg subcutaneously) and placed on a heated surgical table. A midline incision was made, and the left kidney was exposed and blood supply was interrupted by application of microvascular clamps on the renal artery. After 30 min, the clamps were released, and reperfusion was visualized. The right kidney was then exposed, and ligation of ureter and renal blood vessels with a 4-0 suture was performed before right kidney nephrectomy. After surgery, analgesic was administered (dextrose 2.5%, 0.3 ml).

Murine ischemic hindlimb model

Unilateral hindlimb ischemia, a model of persistent vascular injury (76), was surgically induced in mice by femoral arteriectomy as previously described (77). Briefly, the animals were anesthetized with 2% isoflurane, after which an incision was made in the skin overlying the middle portion of the left hindlimb. After ligation of the proximal end of the femoral artery, the distal portion of the saphenous artery was ligated, and the artery and all side branches were dissected free and excised. The skin was closed with a prolene monofilament (6-0) (Johnson & Johnson).

Isolation of apoptotic exosome-like vesicles from murine serum

Serum samples collected from mice before surgery or 2 days after surgery [50 μl (for proteasome activity assay) or 25 μl (for Western blot analyses)] were diluted in 12 ml of RPMI and ultracentrifuged first at 50,000g for 15 min at 4°C to pellet cell debris, apoptotic bodies, and microparticles and then at 200,000g for 18 hours at 4°C to pellet exosome-like vesicles. After the last centrifugation, the pellets were resuspended in 2.4 μl/μl of original serum sample in either PBS (for proteasome activity assay) or Laemmli buffer (for Western blot analyses).

Injection of murine apoptotic endothelial membrane vesicles

Serum-free media conditioned by 1 × 104 mECs were fractionated using sequential centrifugation: a first centrifugation 1200g for 15 min at 4°C to pellet cell debris; a second centrifugation at 50,000g for 15 min at 4°C to pellet apoptotic bodies; and a final ultracentrifugation at 200,000g for 18 hours at 4°C to pellet exosome-like vesicles. Pellets containing either apoptotic bodies or exosome-like vesicles were resuspended in half of the initial volume of conditioned medium. Nongrafted or transplanted mice received tail vein intravenous injections of resuspended preparations (150 μl) every other day during 3 weeks for a total of eight doses. In transplanted mice, the first injection occurred 2 days after transplantation.

Assessment of circulating levels of total IgGs, ANAs, and anti-LG3

Antinuclear antibodies and total IgG levels were assessed using ANA, Mouse, BioAssay kits (US Biologicals) and Mouse IgG total Ready-SET-Go kits (Affymetrix), respectively, according to the manufacturers’ instructions. Anti-LG3 titers were measured with a locally developed enzyme-linked immunosorbent assay. Recombinant LG3 was produced and purified as previously described (78). The purity of the recovered LG3 protein was assessed by reducing SDS–polyacrylamide gel electrophoresis and Coomassie blue R250 staining. Recombinant mouse LG3 (10 ng/μl) was first coated on 96-well Immulon II HB plates (Thermo Electron), for a total of 1 μg per well. Notably, mouse and human LG3 fragments were highly (87%) homologous at the amino acid level. The sera were diluted (1:250), and 100 μl was added per well. The plates were washed, and bound IgGs were detected using horseradish peroxidase coupled with anti-human or anti-mouse IgG (Amersham and Santa Cruz Biotechnology, respectively) or IgG1, IgG2a, IgG2b, and IgG3 antibodies for IgG subclass studies (Santa Cruz Biotechnology). Reactions were revealed with 100 μl of tetramethylbenzidine substrate (BD Biosciences) and stopped with 50 μl of sulfuric acid (H2SO4). Spectrophotometric analysis was taken at 450 nm, and the results were expressed as optical density.

Measurement of murine anti–donor IgG

Sera were diluted 1:100 in fluorescence-activated cell sorting (FACS) buffer and incubated with 1 × 106 BALB/c splenocyte targets for 30 min at 4°C. The samples were then washed three times and stained with phycoerythrin (PE) goat anti-mouse IgG (1:100; Alexa 488 anti-mouse CD3e, BD Biosciences) in FACS buffer for 30 min at 4°C in the dark. Samples were run on a flow cytometer (FACScan, BD) and analyzed on computer software (FACSDiva, BD). CD3+ parent gate was used to avoid nonspecific background signal due to Fc receptor–expressing cells.

Flow cytometric analyses of splenic germinal center B cells and follicular helper T cells

Single-cell suspensions from the spleen were prepared and stained immediately for three-color flow cytometry. The percentage of CD4+PD-1+CXCR5+ follicular helper T cells among CD4+ cells was determined by flow cytometry using anti–programmed death-1 (PD-1) fluorescein isothiocyanate (eBioscience), anti-CXCR5 biotin (BD Biosciences), anti-CD8 allophycocyanin (APC) (BioLegend), anti-CD4 peridinin chlorophyll protein (BD Biosciences), anti–CD62 ligand PE (BD Biosciences), and anti-CD44 Pacific blue (BioLegend) monoclonal antibodies (mAbs) and strep APC Cy7 (BD Biosciences). The percentage of B220+GL7+Fas+ germinal center B cells among B220+ B cells was determined by flow cytometry with anti-B220 APC Cy7 (BD Biosciences), anti-GL7 FITC (BD Biosciences), and anti-Fas PE (eBioscience) mAbs.

Immunohistochemistry

Transplanted and adjacent native aortas were harvested 3 weeks after transplantation. Tissues were fixed with 10% neutral buffered formalin and paraffin-embedded according to established methods. Samples were cut into 4-μm slices. Immunohistochemistry was assessed on an immunostainer (Discovery XT system, Ventana Medical Systems) according to the manufacturer’s recommendations. Antigen retrieval was performed with proprietary reagents. When mentioned, samples were stained with H&E. For the detection of CD3+ and F4/80+ cells and C4d deposition, indirect immunoperoxidase staining was performed using the primary anti-CD3 or anti-F4/80 antibody (1:50; AbD Serotec) or anti-C4d (1:50; Biomedica), respectively, followed by incubation with specific secondary biotinylated antibodies. Streptavidin horseradish peroxidase and 3,3′-diaminobenzidine were used according to the manufacturer’s instructions (DABMap detection kit, Ventana Medical Systems). Finally, sections were counterstained with hematoxylin. Digital images of tissues were captured by Leica DMLS microscope and Leica DFC420C camera (Leica Microsystems). Intimal and medial area grafts were outlined and quantified using a digital image analysis program (ImageJ 1.42q, National Institutes of Health). The T cells and macrophages in the tissue were quantified by determining the number of CD3+ or F4/80+ cells, respectively, per six high-power fields per allograft. C4d deposition was quantified using Visiomorph VIS Histoinformatics Software (Olympus)

Statistical analyses

All data are expressed as means ± SEM derived from at least three independent experiments unless otherwise specified. Biological and histological data were compared using t test. Statistical analyses were performed using Prism 4 (GraphPad software Inc.). P values <0.05 were considered significant.

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/7/318/318ra200/DC1

Fig. S1. Serum deprivation significantly increased caspase-3 activation and the percentage of cells with chromatin condensation in the absence of cell membrane permeabilization.

Fig. S2. Small-particle flow cytometry size calibration.

Fig. S3. Characterization of small particles secreted by apoptotic ECs sensitive to detergent treatment.

Fig. S4. Inhibition of autophagy with bafilomycin does not modulate exosome-like vesicle secretion by serum-starved ECs.

Fig. S5. Electron micrographs of apoptotic bodies released by serum-starved HUVECs and isolated by sequential ultracentrifugation.

Fig. S6. Electron micrographs of exosome-like vesicles released by serum-starved HUVECs and isolated by sequential ultracentrifugation.

Fig. S7. Gene ontology analysis of cellular components and biological processes for proteins unique to exosome-like vesicles and apoptotic bodies.

Fig. S8. Characterization of the presence of classical exosomal markers in the apoptotic exosome-like vesicles proteome.

Fig. S9. Redundant peptide count of perlecan in exosome-like vesicles and apoptotic bodies.

Fig. S10. Injecting equal amounts of proteins from preparations of apoptotic bodies and preparations of apoptotic exosome-like vesicles fail to unmask immunogenic activity in apoptotic bodies.

Fig. S11. Apoptotic exosome-like vesicles, unlike apoptotic bodies, favor B cell responses and autoimmunity.

Fig. S12. Intima/media ratio in murine allografts 3, 6, or 9 weeks after aortic transplantation.

Fig. S13. Characterization of anti-LG3 IgG subclasses.

Fig. S14. Healthy ECs secrete exosome-size vesicles lacking caspase-like proteasome activity.

Fig. S15. Apoptotic vascular smooth muscle cells and tubular epithelial cells also secrete active 20S proteasome in exosome-size EVs.

Fig. S16. Bortezomib treatment does not affect cell death levels.

Fig. S17. The proteasome is not detected in serum membrane vesicles isolated before surgery.

Table S1. Quantitative proteomic identifications based on spectral counts in nanovesicles and apoptotic bodies from serum-starved apoptotic HUVECs.

Table S2. Quantitative proteomic identifications based on spectral counts in nanovesicles from serum-starved apoptotic HUVECs treated with the proteasome inhibitor bortezomib or vehicle.

Table S3. Ubiquitylated proteins identification by MS in nanovesicles from serum-starved HUVECs treated with the proteasome inhibitor bortezomib or vehicle.

Data S1. Source data, unedited gels.

Data S2. Source data, flow cytometry gating.

REFERENCES AND NOTES

  1. Acknowledgments: We wish to thank M. Hénault-Rondeau, S. Morissette, and F. Marsan for their work with the Université de Montréal Renal Transplant Biobank; Y. Durocher for providing recombinant LG3, Institute for Research in Immunology and Cancer (IRIC) histology platform, and CRCHUM cytometry and microscopy platforms. Funding: We acknowledge support from the Canadian National Transplantation Research Program (CNTRP) (M.D., J.-F.C., C.P., E.B., P.T., and M.-J.H.), the Canadian Institutes of Health Research (CIHR) (MOP-15447) (M.-J.H.), the Kidney Foundation of Canada (M.-J.H.), Canadian Cancer Society Research Institute (CCSRI) grant (P.T.), Natural Sciences and Engineering Research Council of Canada (NSERC) grant 311598 (P.T.), and NSERC grant 386598 (E.B.). M.-J.H. is the holder of the Shire Chair in Nephrology, Transplantation and Renal Regeneration of Université de Montréal. We thank the J.-L. Lévesque Foundation for renewed support. IRIC receives infrastructure support from the Canadian Center of Excellence in Commercialization and Research, the Canadian Foundation for Innovation, and the Fonds de recherche du Québec—Santé (FRQS). E.B. is a recipient of a Canadian Institutes of Health Research new investigator award. M.R. was supported by a grant from NSERC. N.P. received a fellowship grant from the University of Montreal Nephrology Consortium. Author contributions: M.D., C. Bell, E.B., P.T., and M.-J.H. developed concepts, designed the experiments, and wrote the manuscript. M.D., C. Bell, J.T., D.B., L.P., B.Y., K.H., S.Q., A.L., N.P., C. Béland, A.-C.D., T.L., W.D., M.R., C.R., D.G., and E.B. performed the experiments and analyzed the data. H.C. was the consultant for the statistical analyses. J.-F.C., M.R., A.R., D.M., H.C., C.P., M.D., and E.B. analyzed the data, gave technical support and conceptual advices, and edited the manuscript. Data and materials availability: The proteomic data for this study are deposited in ExoCarta and vesiclepedia databases (unique accession id: vesiclepedia_560).
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article