Published online before print November 1, 2007
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
* National Heart and Lung Institute, Imperial College at Harefield Hospital, and
Transplant Immunology, Harefield Hospital, Harefield, Middlesex, United Kingdom;
Eric Bywaters Centre, BHF Cardiovascular Medicine Unit, Imperial College at Hammersmith Hospital, London, United Kingdom; and
James Hogg iCAPTURE Centre for Pulmonary and Cardiovascular Research, St. Paul’s Hospital, University of British Columbia, Vancouver, British Columbia, Canada
1 Correspondence: National Heart and Lung Institute, Heart Science Centre, Harefield Hospital, Harefield, Middlesex, UB9 6JH, UK. E-mail: marlene.rose{at}imperial.ac.uk
ABSTRACT
Anti-vimentin antibodies (AVA) are associated with autoimmunity and solid organ transplantation, conditions associated with vascular disease, but their contribution to disease pathogenesis is unknown. Here, we have examined interactions between AVA (mAb and serum from patients) and various leukocyte populations using whole blood and flow cytometry. Normal blood treated with patient sera containing high AVA-IgM titers or with a vimentin-specific monoclonal IgM led to activation of platelets and other leukocytes, as demonstrated by induced expression of P-selectin, fibrinogen, tissue factor, and formation of platelet:leukocyte (P:L) conjugates and a reduction in platelet counts. This activity was antigen (vimentin)-specific and was not mediated by irrelevant IgM antibodies. Flow cytometry demonstrated that AVA do not bind directly to resting platelets in whole blood, but they bind to 
10% of leukocytes. Supernatant, derived from AVA-treated leukocytes, induced platelet activation, as measured by the generation of platelet microparticles, when added to platelet-rich plasma. When AVA were added to whole blood in the presence of CV-6209, a platelet-activating factor (PAF) receptor inhibitor, platelet depletion was inhibited. This suggests that PAF is one of the mediators released from AVA-activated leukocytes that leads to P:L conjugation formation and platelet activation. In summary, AVA bind to leukocytes, resulting in release of a PAF and prothrombotic factor that exert a paracrine-activating effect on platelets. Overall, this proposed mechanism may explain the pathogenesis of thrombotic events in autoimmune diseases associated with AVA.
Key Words: neutrophils • senescence • complement
INTRODUCTION
Autoantibodies to the intermediate filament vimentin are associated with rheumatoid arthritis [1 ], systemic lupus erythematosus (SLE) [2 3 4 ], and rejection of solid organ transplants [5 6 7 8 9 ]. Vimentin is a cytoskeleton intermediate filament protein present in cells of mesenchymal origin; this includes leukocytes, endothelial cells, and smooth muscle cells. More recently, it has been observed on the cell surface of apoptotic cells [10 11 12 13 ] and thrombin-activated platelets [14 ]. The production of anti-vimentin antibodies (AVA) in certain diseases is probably caused by excessive exposure to vimentin on apoptotic cells, as it is known that caspase-dependent cleavage of vimentin, with exposure of vimentin on the cell surface, is a necessary requisite for apoptosis [15 ]. The function of AVA is unknown and in particular, whether they have an active role in disease pathogenesis.
The prevalence of coronary artery disease in SLE patients is high, and heightened states of immune activation and prothrombotic activity are thought to be key contributory factors to disease pathogenesis in these patients [16 ]. Lupus autoantibodies have been implicated, as observed by immune deposits present in kidney and heart [17 ]. In the context of transplantation, AVA have been associated with a different type of atherosclerosis, namely, graft vascular disease (GVD) [9 ], which is the most common complication following heart or renal transplantation; it is characterized by intimal occlusion and fibrosis of donor arteries and veins [18 , 19 ].
A procoagulant microvasculature is associated with the pathogenesis of GVD. The presence of fibrin deposition [20 ] and depletion of tissue plasminogen activator [21 ] in the microvessels of the heart are predictive of heart transplant recipients who will develop GVD. Similarly, deposition of the complement component C4d within allografts is characteristic of GVD [22 ]. These studies suggest that synergy between thrombotic events and complement-fixing antibodies contributes to GVD and possibly, to atherosclerosis in patients with autoimmune diseases. We hypothesize that antibodies to the autoantigen vimentin, particularly of the IgM subclass, may interact with vimentin-expressing platelets with possible pathogenic consequences. This hypothesis was tested in vitro by adding AVA to normal whole blood, platelet-rich plasma (PRP), and leukocyte-rich plasma and investigating its effect on formation of platelet:leukocyte (P:L) conjugates and platelet microparticles (PMP).
MATERIALS AND METHODS
Blood collection and patient serum
MHC HLA-A2 subtype individuals consented to provide normal whole blood, which was collected into 4.5 mL citrate-theophylline-adenosine-dipyridamole tubes and then diluted 1/10x in Tyrodes buffer. PRP was obtained by collecting the supernatant fraction of blood centrifuged at 200 g for 10 min. Post-transplant serum collected from cardiac transplant recipients for diagnostic reasons was used with permission from our local ethics committee. Patient serum had been screened for the presence of anti-vimentin and HLA antibodies as described previously [9
, 23
].
Preparation of recombinant human vimentin
The cDNA for human vimentin was isolated from a HUVEC cDNA expression library by PCR, introducing a 5' NdeI restriction site and a BamHI site at the 3' end of the cDNA (primers: forward 5' ATA GAG CAT ATG TCC ACC AGG TCC GTG TCC; reverse 5' GCG CTC GGA TCC TCT TAT TCA AGG TCA TCG TG). The PCR product was subcloned into NdeI/BamHI of pET15b (Novagen, Merck Biosciences, Nottingham, UK), a bacterial expression vector, and transformed into BL21 Escherichia coli (BL21/vimentin, Novagen). Crude preparations of recombinant human vimentin were prepared and extracted according to the pET system manual (Novagen) and purified on a His-Bind resin column under denaturing conditions, using a His-Bind purification kit (Novagen). Purification was confirmed by SDS-PAGE (a single band at 58 kD was observed) and mass spectrometry (not shown).
Depletion of AVA from patient sera
Recombinant human vimentin protein (1 mg/mL in 6 M urea, preparation) was conjugated to agarose beads as described in the kit instructions (AminoLink Plus immobilization kit). Vimentin-conjugated agarose beads (50 µL) were incubated with 50 µL patient serum at 25°C overnight and then centrifuged at 1000 g for 10 min, and the supernatant fraction, depleted patient serum, was collected. ELISA assays [10
] demonstrated that this treatment reduces AVA titers (mean nondepleted, 1061±52.9; mean depleted, 160±29.3).
Flow cytometry and mAb
Equal amounts (25 µg/mL final concentration) of mAb mouse IgM antibody to HLA-A2 and -A3, from One Lambda (Canoga Park, CA, USA), and AVA 13.2 IgM and V9-IgG (both mouse anti-human, Sigma Chemical Co., St. Louis, MO, USA) were added to 20 µL-diluted (1/10x) whole blood and incubated at 37°C for 30 or 45 min. Patient sera (20 µL) were added to 20 µL diluted blood for 30 or 45 min. CD41-retinal pigment epithelial (RPE), C3d-FITC, or fibrinogen-FITC antibodies (mouse anti-human, rabbit anti-human, rabbit anti-human, respectively, all from BD Biosciences, San Jose, CA, USA), at a final concentration of 25 µg/ml, were used to label each sample to detect platelets and expression of C3d or fibrin. CD62P-allophycocyanin (APC; mouse IgG1 anti-human, BD Biosciences) and tissue factor-FITC antibodies (rabbit anti-human, American Diagnostica, Stamford, CT, USA) were also added to blood at similar concentrations. Isotype controls, FITC rabbit IgG (from Beckman Coulter, Fullerton, CA, USA), APC-mouse IgG1 (from Beckman Coulter), and RPE-mouse IgG1 (from BD Biosciences) were added at the same concentrations. To label leukocytes, Hoechst 33342 (Molecular Probes, Eugene, OR, USA), diluted in PBS (1 µg/L final concentration), was subsequently added for another 30 min. Labeled leukocytes were identified as "Hoechst+ve cells" during flow cytometric analysis. To track binding of mAb IgM antibodies in whole blood, FITC-goat anti-mouse IgM (Sigma Chemical Co.) was added at a final concentration of 25 ug/ml. To determine numbers of vimentin-positive cells in whole blood, 13.2 IgM was added to washed buffy coat leukocytes, which were resuspended in 100 ul PBS, to which was added 13.2 IgM (final concentration, 25 ug/ml). After 30 min incubation at 4°C, cells were washed again, and FITC-goat anti-mouse IgM was added (final concentration, 10 ug/ml) for another 30 min at 4°C. Cells were analyzed using a Becton Dickinson FACS-Aria instrument with 405, 488, and 635 nm single line lasers, and >30,000 events were analyzed for every sample.
Measurement of PMP was performed using 1.0 µm beads from Invitrogen (Carlsbad, CA, USA; Cat. #F13080). The platelet activating factor (PAF) inhibitor CV-6209 was obtained from Calbiochem (Nottingham, UK).
Complement-dependent cytotoxic assay on AVA-treated, purified leukocytes
To prevent contamination by platelets, a modified method of purifying leukocytes was performed. Normal blood was collected into acid citrate dextrose tubes (BD Biosciences), and leukocyte-rich preparation was prepared by centrifugation of 10 mL whole blood for 10 min at 200 g, and the PRP supernatant was removed. An additional 1.0 mL Tyrodes buffer was added to the remaining whole blood, mixed by inversion, and then centrifuged again under the same conditions. The supernatant was removed again, and remaining blood was lightly layered onto Lympholyte H solution (Cedarlane Labs, Hornby, ON, Canada) in a 15-mL Falcon tube. This was centrifuged at 200 g for 20 min. Upon centrifugation, four layers were present from top to bottom. A Pasteur pipette was used to suction out the leukocyte layer (typically 2 mL from a 10-mL whole blood preparation). This layer was centrifuged at 200 g for 5 min to remove residual platelets. The supernatant containing platelets was discarded, and the pellet was resuspended in modified Tyrodes buffer. Flow cytometry demonstrated these leukocytes to be 60% neutrophils; the remainder was monocytes and lymphocytes, and they were used for two tests: the cytotoxic assay and preparation of supernatants from AVA-activated neutrophils.
Leukocytes were diluted to 5 x 105 cells/µL, and modified Tyrodes buffer and 1.0 µL aliquots were placed in single wells of a Terasaki plate. To these were added 1 µL HLA-A2 IgM, HLA-A3 IgM, or 13.2 IgM antibody (0.5 µg/mL) in each well in duplicate. After 30 min, 5 µL rabbit complement (Cedarlane Labs) was added and incubated for another 30 min. Cell viability was assessed by adding 2 µL FluoroQuench (One Lambda) to each well. A Zeiss inverted microscope fitted with a QImaging Retiga EXi color-cooled camera and QCapture Pro software (QImaging Inc., B.C., Canada) was used to acquire images. A live/dead filter set (Chroma Technology Corp., Germany) was used to visualize the ethidium bromide and acridine orange viability stains.
RESULTS
Effect of monoclonal and patient AVA on whole blood
Treatment of whole blood with AVA 13.2 IgM resulted in a depletion of platelet counts (determined by numbers of CD41+ve cells) as well as an increase in percent P:L conjugates (calculated as a percentage of CD41+Hoechst+ve cells/Hoechst+ve cells) compared with whole blood treated with AVA V9 IgG antibody or not (Fig. 1
, A vs. B and C). The effect of AVA 13.2 IgM was inhibited in the presence of recombinant human vimentin (Fig. 1D)
.
 View larger version (23K): [in a new window] |
Figure 1. AVA mAb induce P:L conjugate formation and surface expression of C3d and fibrinogen. (A–D) Forward/side-scatter (FSC and SSC, respectively) plots of blood under different treatments incubated for 30 min. The green population represents platelets (CD41+ve); the red represents nucleated leukocytes (Hoescht+ve). Incubation with 13.2 IgM (A) resulted in platelet depletion and P:L conjugate formation (assessed by CD41+ve cells/Hoescht+ve cells), whereas incubation with V9 IgG (B) gave similar results to untreated (C) blood. P:L conjugate formation induced by 13.2 IgM was inhibited by recombinant vimentin protein + 13.2 IgM (D). (E) Data about the percentage of total platelets and leukocytes expressing fibrinogen (open bars) and C3d (solid bars) are summarized, before and after AVA 13.2 IgM treatment. Representative of five experiments. *, P < 0.05; t-test, compared with untreated. (F and G) Demonstrate that 13.2 IgM-activated cells do not bind nonspecific FITC-rabbit IgG.
|
To compare the activity of the AVA monoclonal with AVA from transplant patient sera, normal blood was then treated with patient sera containing high titer IgM AVA (mean titer, 1061±52.9) or the same sera depleted of AVA (mean titer, 160 ±29.3) using vimentin-coated agarose beads. These sera did not contain alloantibodies or AVA of the IgG subclass. Patient sera containing high AVA titer lead to a decrease in the ratio of free P:L compared with whole blood treated with control serum or sera depleted of AVA IgM (Fig. 2A ). In Figure 2B , an increase in P:L conjugates was observed when whole blood was treated with patient sera with high AVA IgM compared with normal blood incubated with control serum (from a transplant patient negative for AVA) or sera depleted of AVA IgM. To control for the possibility that vimentin-coated beads may remove Ig nonspecifically, we treated patient serum known to contain high titers of antibodies to HLA with vimentin-coated agarose beads; this had no effect on the HLA antibody titer (not shown). These results demonstrate that the AVA in patient’s sera can also induce formation of P:L conjugates.
 View larger version (16K): [in a new window] |
Figure 2. Patient sera with AVA induce platelet activation and formation of P:L conjugates. Normal blood was treated with patient sera containing high titer AVA (solid bars), control serum (labeled control patient), or the same patient sera depleted of AVA using agarose-coated beads (open bars) for 30 min. (A) The extent of platelet activation was characterized by the ratio of free P (CD41+ve cells):L (Hoechst+ve cells). (B) Formation of P:L conjugates was characterized by the percentage of cells coexpressing Hoechst + CD41 over the total number of Hoechst+ve cells.
|
 View larger version (44K): [in a new window] |
Figure 3. AVA and HLA IgM induce P:L conjugates, which express tissue factor and P-selectin. Whole blood was treated with AVA 13.2 (A), HLA-A2 IgM (B), or HLA-A3 IgM (C) or untreated (D). The effects of IgM treatment on platelets and leukocytes were expressed as forward/side-scatter plots in A–D (the blue area indicates P:L conjugates). (E) The extent of P:L conjugate formation is summarized in all four groups. (F) The extent of tissue factor(+)ve (TF; open bars) and P-selectin(+)ve (solid bars) P:L conjugates is summarized in all four groups. Representative of two independent sets of four experiments. *, P < 0.05; t-test, compared with untreated. (G and H) HLA-A2 IgM-treated blood cells do not bind nonspecific FITC rabbit IgG (G) or APC-mouse IgG1 (H).
|
11% of CD41-PE-positive particles in the sample were of microparticle size and morphology. When this plasma was treated with AVA 13.2 IgM, the percent PMP at 4.8 ± 0.8% was lower than the percent PMP observed in PRP. Treatment with HLA-A2 IgM resulted in 40.0 ± 7.5% PMP, presumably a result of complement-mediated lysis by the HLA-A2 IgM [24
]. Treatment of PRP with the HLA-A3 IgM resulted in 7.9 ± 1.3%, a level similar to untreated PRP. These observations strongly suggest that the 13.2 IgM, unlike HLA-A2 IgM, has a negligible effect on resting platelets, which are negative for cell surface vimentin but constitutively express MHC class I antigens. Overall, the results suggest that the platelet depletion observed in the whole blood and PRP by AVA-IgM was caused by an initial interaction of AVA-IgM with nonplatelet leukocytes, which ultimately leads to platelet activation and vimentin surface expression on platelets.
 View larger version (23K): [in a new window] |
Figure 4. AVA-IgM does not induce platelet activation directly. (A–D) Forward/side-scatter plots in a density contour format from a single representative experiment. Platelet activation is expressed as "% PMP"; this was calculated as number of CD41+ve particles with a diameter below 1.0 µm over the total number of CD41+ve events. Results of %PMP after treatment of PRP with 13.2 IgM (B), HLA-A2 IgM (C), and HLA-A3 IgM (D) compared with untreated (A) for 30 min are summarized in E. Representative of four experiments. *, P < 0.05; t-test, compared with normal group.
|
15% of leukocytes bound AVA 13.2 IgM. These vimentin-positive cells were predominantly neutrophils, as determined by their light-scattering properties. Hence, a subpopulation of leukocytes, expressing vimentin, is the binding site for AVA-IgM antibodies.
 View larger version (26K): [in a new window] |
Figure 5. Localization of IgM to granulocytes and activated platelets and their cytotoxic effect on leukocytes. FITC goat anti-mouse IgM was added to whole blood to track HLA-A2 IgM (A), HLA-A3 IgM (B), and AVA 13.2 IgM (C) or untreated blood (D). The white population denotes cells with positive IgM binding. Representative of three experiments. (E) Flow cytometry of buffy coat leukocytes to which was added AVA 13.2 IgM (red line) or HLA-A3 IgM (gray line), followed by FITC-goat anti-mouse IgM. Representative of three experiments. (F) HLA-A2 IgM, HLA-A3 IgM, and AVA 13.2 IgM were added to purified leukocytes and incubated with excess complement to induce antibody-mediated cell death. Viability was assessed by ethidium bromide/acridine orange labeling of treated cells. Representative of three experiments.
|
Supernatant of AVA-activated leukocytes induces platelet activation
To determine if AVA-bound leukocytes release mediators that subsequently activate platelets, leukocytes were purified and treated with AVA 13.2 IgM, HLA-A2 IgM, or HLA-A3 IgM for 30 min, and the supernatant was transferred to purified platelets (PRP). Platelet activation was assessed by generation of PMP at 30 min. Supernatant from HLA-A3 IgM did not result in significant PMP formation (Fig. 6B
) compared with untreated control (Fig. 6A)
. Supernatant from leukocytes treated with AVA 13.2-IgM resulted in substantial PMP formation (Fig. 6C)
. Quantitative data from these experiments are presented in Figure 6E
. The large amount of PMP generated by transfer of supernatant from HLA-A2 IgM-treated leukocytes is probably partially a result of carryover of the HLA-A2 IgM antibody, which causes direct activation of HLA-A2-positive, resting platelets (Fig. 4C)
. Although the supernatant from AVA 13.2 IgM-treated leukocytes contains released inflammatory mediators and the AVA IgM, we demonstrated previously (Fig. 4B
and 4E)
that the AVA IgM alone does not have an activating effect on platelets. Hence, the activation observed in Figure 6C
indicates activation is a result of the released inflammatory mediators and not the AVA 13.2 IgM.
 View larger version (25K): [in a new window] |
Figure 6. Effect of supernatant (SN) from IgM-treated leukocytes on platelet activation. Results are expressed as an analysis of platelet particles according to size, and only CD41+ particles are presented above. Particles in black are CD41+, and particles in white are CD41+ cells with a forward-scatter lower than a 1.0-µm diameter fluorescent-counting bead, a region of particles that is considered to be PMP. (A) Untreated platelets. (B) Supernatant from HLA-A3 IgM treatment did not result in PMP formation greater than control (A). Treatment of PRP with supernatant from AVA-IgM resulted in platelet aggregation and PMP formation (C), and supernatant from HLA-A2 IgM resulted in the highest generation of PMP compared with all treatments (D). (E) Quantitative data are shown. Representative of three experiments. *, P < 0.05; t-test, compared with no treatment.
|
 View larger version (36K): [in a new window] |
Figure 7. PAF inhibition attenuates platelet activation and blood cell agglutination. Whole blood was untreated (A) or incubated for 45 min with 13.2 IgM AVA in the absence (B) or presence (C) of the PAFR inhibitor CV-6209. Flow cytometry of treated blood was used to determine platelet depletion (D) and C3d expression on all platelets and leukocytes (E; where open bars represent leukocytes, and solid bars represent platelets). The white-colored particles in scatterplots (A–C) represent platelets with C3d(+)ve signal, and the position of platelets is marked as a bracket in A. Pretreatment of whole blood with CV-6209 attenuated platelet depletion, but an increase in C3d deposition on platelets was still observed (A vs. C). Platelet counts were determined by analyzing 5000 leukocyte events for each sample. Representative of four experiments. *, P < 0.05; t-test, compared with no treatment.
|
AVA have been described in a number of diverse conditions including autoimmunity [1 , 2 , 25 ], chronic infections [26 ], and clinical rejection of solid organ allografts [5 6 7 8 9 ]. It has been demonstrated recently that vimentin-immunized mice undergo accelerated rejection of their cardiac allografts [27 ], but the mechanisms were not elucidated. This is the first study to demonstrate interactions between IgM AVA and leukocytes and to suggest a mechanism for their pathogenesis. The study was initiated by the observation that activated platelets express cell surface vimentin [14 ], leading us to hypothesize that AVA would have an effect on platelet activation or thrombosis. Treatment of normal whole blood with the AVA 13.2 IgM monoclonal resulted in P:L conjugate formation and depletion of platelet counts. This was accompanied by induction of P-selectin on platelets attached to leukocytes (Fig. 3F) and deposition of fibrinogen, C3d, and tissue factor on platelets and leukocytes. P-selectin is known to be a crucial molecule in P:L conjugate formation [28 ]. That these effects can be produced by patients’ antibodies was demonstrated by use of serum from cardiac transplant recipients, which were selected on the basis of their high IgM AVA titer and absence of alloantibodies; when these serum samples were depleted of AVAs, the formation of P:L conjugates was prevented, and platelet counts were unaffected.
To compare the effects of the AVA IgM to other antibodies known to bind to platelets, IgM specific for HLA antigens were chosen. Antibodies to HLA-A2 but not HLA-A3 had an effect on blood from a HLA-A2-positive individual, confirming the importance of the antigen-binding part of the IgM molecule and discounting the possibility that IgM antibodies could activate platelets nonspecifically to form P:L conjugates. Experiments using purified platelets as opposed to whole blood described an important difference between HLA-A2 IgM and AVA IgM, namely, that only anti-HLA antibodies bind to purified platelets, and AVA IgM did not. This is not surprising, as it is known that only activated and not quiescence platelets express surface vimentin [14 ]. Podor et al. [14 ] also described vimentin expression on PMP; this will explain the decrease in percent PMP observed when AVA 13.2 IgM was added to purified platelets (Fig. 4A and 4B) . We suggest that AVA IgM binds to the vimentin-positive PMP in normal PRP, resulting in complement-mediated lysis and further degradation of PMP.
Using whole blood and buffy coat leukocytes, it was determined that the 13.2-IgM AVA binds to 15% of circulating leukocytes, defined as neutrophils and monocytes by their light-scattering properties. Moisan and Girard [29
] described a similar number of vimentin-positive neutrophils in normal blood and demonstrated them to be spontaneously apoptosing neutrophils. It is established that apoptotic cells demonstrate surface expression of vimentin [10
, 12
, 13
]; conversely, recent papers have described expression of vimentin on the surface of nonapoptotic cells [30
31
32
], raising the possibility that AVA may interact with viable leukocytes in whole blood. The fact that AVA V9 IgG antibody failed to activate formation of P:L conjugates is interesting, as this mAb recognizes the C-terminus of the vimentin moiety [14
, 29
], which is not expressed on viable neutrophils [29
], leukemic cell lines, or activated T cells [30
] but is expressed on leukocytes undergoing apoptosis [29
]. Our own studies indicate that the V9 antibody binds to 5–10% of leukocytes in untreated whole blood (not shown). The failure to activate in this study may reflect a relative lack of exposure of the epitope recognized by V9 or lack of complement-fixing ability of the IgG molecule compared with IgM antibodies, or alternatively, it may be that living cells are necessary to transduce cell surface vimentin-mediated activation signals. Further studies are in progress to investigate the viability, phenotype, and morphology of the AVA-binding leukocytes in this study and the effect of IgG AVA.
The possibility that AVA bind to leukocytes that subsequently release factors to activate resting platelets was confirmed by using supernatant from AVA-treated leukocytes and adding it to platelets and observing PMP formation (Fig. 6) . We hypothesied that PAF was the most likely factor implicated in this effect, as it can be synthesized rapidly and released by activated leukocytes, which include neutrophils [33 ] and monocytes [34 , 35 ]. In our studies, leukocytes expressed tissue factor following AVA treatment of whole blood (Fig. 3F) . This is also likely to have originated from activation of leukocytes [36 , 37 ] by AVA; however, unlike PAF, which has direct platelet-agonist effects, tissue factor does not activate platelets directly [38 ]. It is known that PAF binds to PAFR present on platelets and leukocytes [39 ]. Upon PAF binding, calcium channels are opened, initiating activation of the platelet. The addition of the PAF inhibitor to whole blood before addition of 13.2 AVA had an interesting effect on platelets (Fig. 7) ; although it inhibited depletion of free platelets, the platelets were still expressing C3d (Fig. 7E) . This demonstrates platelet activation without lysis. It appears that the PAF inhibitor did not affect release of activating factors from the leukocytes but inhibited opening of the calcium channels necessary for cell lysis. It is unlikely that PAF is the only mediator released by AVA-activated leukocytes. In summary, we hypothesize that AVA induce platelet activation and PMP formation via four stages: activation of leukocytes and release of PAFs and expression of tissue factor; induction of P-selectin, vimentin, and tissue factor on platelets; binding of fibrinogen to activated platelets (via GPIIbIIIa) and formation of P:L conjugates; and binding of AVA to activated platelets and generation of PMP. The latter process is likely to be mediated by complement, as demonstrated by the presence of C3d on platelets and the ability of AVA to fix complement and cause leukocyte lysis in vitro (Fig. 5C) . Previous studies have shown sensitivity of platelets to complement-mediated lysis [24 , 40 ]. We have not yet formally demonstrated the complement dependence of this process using complement inhibitors.
The formation of P:L conjugates by AVA is an observation that sheds light on possible hemostatic mechanisms leading to GVD development in allografts. It may also be important for atherosclerotic disease progression in autoimmune diseases such as SLE, which are characterized by AVA. When P:L conjugates are formed, the effectiveness of these leukocytes to roll and tether to activated endothelium is increased substantially [41 , 42 ]. The release of tissue factor or its expression on activated leukocytes may also potentiate T cell activation [43 ], alongside its procoagulation effects.
This study has described effects of IgM AVA on leukocytes, which are antigen-dependent. In 1984, Hansson et al. [44 ] demonstrated Fc-dependent binding of nonspecific IgG to vimentin exposed on the surface of damaged endothelial cells; in the current study, the fact that HLA-A3 IgM and HLA-A2 IgM behave in a different way than AVA 13.2 IgM and the fact that vimentin-coated agarose beads deplete only AVA from patients’ serum (and not IgG HLA antibodies) mitigate against vimentin acting as a general FcR for circulating Ig.
It is interesting to speculate that the effects of AVA described here, in vitro, may be partly responsible for the neutropenia and thrombocytopenia typically present in lupus patients. Indeed the effects of AVA on vimentin-positive leukocytes may be analogous to the effect of antineutrophil cytoplasmic autoantibodies, which are associated with specific forms of systemic vasculitis [45 ]. In the latter case, it is known that cytokine treatment exposes these autoantigens, proteinase-3 and myeloperoxidase, on the neutrophil surface, although the in vivo stimulus of such exposure is not known. In conclusion, the mechanism we describe here may reflect a novel mechanism of how autoantibodies lead to thrombosis and atherosclerosis.
ACKNOWLEDGEMENTS
H. S. L. ant T. J. P. were funded by BHF FS/05/124/19972, CIHR IOPG, and CIHR Grant 20R90839. B. M. M. was funded by a BTS Fellowship, and M. L. R. was supported by BHF program grant RG/2001005.
Received June 1, 2007; revised September 24, 2007; accepted September 24, 2007.
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
