Identification of target antigens of anti-endothelial cell and anti-vascular smooth muscle cell antibodies in patients with giant cell arteritis: a proteomic approach

Introduction Immunological studies of giant cell arteritis (GCA) suggest that a triggering antigen of unknown nature could generate a specific immune response. We thus decided to detect autoantibodies directed against endothelial cells (ECs) and vascular smooth muscle cells (VSMCs) in the serum of GCA patients and to identify their target antigens. Methods Sera from 15 GCA patients were tested in 5 pools of 3 patients' sera and compared to a sera pool from 12 healthy controls (HCs). Serum immunoglobulin G (IgG) reactivity was analysed by 2-D electrophoresis and immunoblotting with antigens from human umbilical vein ECs (HUVECs) and mammary artery VSMCs. Target antigens were identified by mass spectrometry. Results Serum IgG from GCA patients recognised 162 ± 3 (mean ± SD) and 100 ± 17 (mean ± SD) protein spots from HUVECs and VSMCs, respectively, and that from HCs recognised 79 and 94 protein spots, respectively. In total, 30 spots from HUVECs and 19 from VSMCs were recognised by at least two-thirds and three-fifths, respectively, of the pools of sera from GCA patients and not by sera from HCs. Among identified proteins, we found vinculin, lamin A/C, voltage-dependent anion-selective channel protein 2, annexin V and other proteins involved in cell energy metabolism and key cellular pathways. Ingenuity pathway analysis revealed that most identified target antigens interacted with growth factor receptor-bound protein 2. Conclusions IgG antibodies to proteins in the proteome of ECs and VSMCs are present in the sera of GCA patients and recognise cellular targets that play key roles in cell biology and maintenance of homeostasis. Their potential pathogenic role remains to be determined.


Introduction
Giant cell arteritis (GCA), also known as temporal arteritis, is a primary systemic vasculitis involving large-and medium-sized vessels. GCA commonly causes bitemporal headaches, jaw claudication, scalp tenderness and/ or abnormal temporal arteries (tender, nodular, swollen and thickened arteries with decreased pulses) detected during physical examinations. GCA does not occur in people younger than 50 years old, and its incidence increases with age and peaks in Caucasians older than 70 years of age [1,2]. Ocular ischaemic complications occur in 25% of the patients and leads to irreversible visual loss in 15% [3]. No definite immunological marker has been identified in GCA, and patients usually present with increased erythrocyte sedimentation rates and/or C-reactive protein levels. Diagnosing GCA can be difficult, and temporal artery biopsy is the gold standard for making the diagnosis [4]. However, in 10% to 20% of patients with GCA, the biopsy shows no specific change [5].
GCA is an inflammatory condition of unknown origin characterised by the presence of giant cells and a remodelling process in the arterial wall [6]. In patients with GCA, an immune-mediated reaction is suspected to be triggered by an antigen of unknown origin, either microbial or a self-antigen, that could be presented to T cells by dendritic cells [7]. Thus, macrophages and giant cells stimulated by interferon-γ (IFN-γ) play a major role in the disruption of the elastic lamina and the remodelling of vessel walls. In addition, in the adventitia, macrophages produce proinflammatory cytokines such as interleukin 1 (IL-1) and IL-6, whereas in the media and intima they contribute to arterial injury by producing metalloproteinases and nitric oxide [6,8,9].
Anti-endothelial cell (anti-EC) antibodies (AECAs) have been detected in a wide range of systemic inflammatory and/or autoimmune diseases, including primary and/or secondary systemic vasculitis [10]. Although the pathogenic role of AECAs remains controversial [11,12], these antibodies may be responsible for EC activation [13] and induction of antibody-dependent, cell-mediated cytotoxicity and apoptosis [14]. In GCA, AECAs were detected in 33% of sera by performing ELISA on fixed human umbilical vein ECs (HUVECs) [15], but their presence was not confirmed by indirect immunofluorescence [16]. Anti-vascular smooth muscle cell (anti-VSMC) antibodies have been detected in an experimental rat model of vasculitis [17]; however, to our knowledge, these antibodies have not been investigated in patients with primary systemic vasculitis.
We used 1-D and 2-D immunoblotting, followed by mass spectrometry (MS), to investigate the presence of autoantibodies directed against ECs and VSMCs and identify their target antigens in patients with GCA.

Patients
Serum samples were obtained from 15 patients who fulfilled the American College of Rheumatology (ACR) criteria for GCA [4] and 33 patients with anti-neutrophil cytoplasm antibody (ANCA)-associated vasculitis who fulfilled the ACR and the Chapel Hill criteria used as vasculitis controls, with the control group comprising 15 patients with Wegener's granulomatosis (WG), 9 with Churg-Strauss syndrome (CSS) and 9 with microscopic polyangiitis (MPA) [18]. In each group of patients with ANCA-associated vasculitis, two-thirds of the patients had active disease as assessed by a Birmingham Vasculitis Activity Score (BVAS) >3 in the absence of treatment, and one-third of the patients had inactive disease as assessed by a BVAS <3. Some patients in both groups either received corticosteroids and/or immunsuppressants at the time of blood sampling. Sera from 12 healthy blood donors were used as healthy controls (HCs). Serum samples were collected from patients and HCs, aliquoted and stored at -80°C until use. Serum samples were used individually for 1-D immunoblotting and pooled for 2-D immunoblotting (five pools of sera from three patients with GCA each, and one pool of sera from twelve HCs). All patients and healthy controls gave their written informed consent to participate in the study. Serum samples were collected with the approval of the ethics committee of the groupe hospitalier Pitié-Salpêtrière, and the study conformed to the principles outlined in the Declaration of Helsinki.

Cell culture
Human internal mammary artery VSMCs were obtained from patients undergoing aortocoronary bypass surgery. All patients gave their written consent, and the protocol for waste surgical tissue was approved by the ethics committee of groupe hospitalier Pitié-Salpêtrière. These cells were immortalised by transduction of a lentiviral vector incorporating the catalytic subunit of the human holoenzyme telomerase RT and T antigen of simian virus 40 in a primary culture of VSMCs as previously described [19]. Immortalised VSMCs were cultured in Smooth Muscle Cell Basal Medium (PromoCell, Heidelberg, Germany) supplemented with decomplemented FCS (5%), insulin (5 μg/mL), basic fibroblast growth factor (bFGF) (2 ng/mL), epidermal growth factor (EGF) (0.5 ng/ml), streptomycin/penicillin (1%) and ciprofloxacin (1%) at 37°C in 5% CO 2 . The VSMC phenotype was confirmed by using smooth muscle myosin heavy chains 1 and 2 and sm22α antibodies (Abcam, Cambridge, UK) (data not shown).

Two-dimensional immunoblotting Protein extracts
HUVECs and VSMCs were stored at -80°C in 1 mM PMSF and protease inhibitors (Complete Mini; Roche Diagnostics, Meylan, France). Protein extraction was performed as described previously [25] (Additional file 1). Briefly, cells were suspended at 1 × 10 6 /mL in a sample solution extraction kit (Kit 3; Bio-Rad Laboratories). Cell samples were sonicated, and the supernatant was collected after ultracentrifugation (Optima L90-K ultracentrifuge; Beckman Coulter, Fullerton, CA, USA) at 150,000 × g for 25 minutes at 4°C. Protein quantification was carried out using the Lowry method [26]. The supernatant was aliquoted and stored at -80°C.

Two-dimensional electrophoresis
Two-dimensional electrophoresis (2-DE), 2-D immunoblotting and protein identification by MS were performed as previously reported [27] and are detailed in Additional file 1.

Modelling with the use of ingenuity pathway analysis software
To gain insight into the biological pathways and networks that were significantly represented in our proteomic data sets, we used ingenuity pathway analysis software (IPA; Ingenuity Systems, Redwood City, CA, USA). IPA selects 'focus proteins' to be used for generating biological networks. Focus proteins are the proteins from data sets that are mapped to corresponding gene objects in the Ingenuity Pathway Knowledgebase (IPKB) and are known to interact with other proteins on the basis of published, peer-reviewed content in the IPKB. From these interactions, IPA builds networks with a size of no more than 35 genes or proteins. A P value for each network is calculated according to the fit of the user's set of significant genes and/or proteins. IPA computes a score for each network from the P value that indicates the likelihood of the focus proteins in a network being found together by chance. We selected only networks scoring ≥ 2 with P < 0.01 of not being generated by chance. Biological functions were assigned to each network by use of annotations from the scientific literature and stored in the IPKB. Fisher's exact test was used to calculate the P value to determine the probability of each biological function and/or disease or pathway being assigned by chance. We used P ≤ 0.05 to select highly significant biological functions and pathways represented in our proteomic data sets. The build function of IPA allows the generation of pathways that can complete the data analysis by showing interactions of identified proteins with a specific group of molecules [28,29].

Results
The clinical and histological characteristics of patients with GCA are summarised in Additional file 2, Supplemental Table S1. The mean age (± SD) of the patients with GCA was 74.8 ± 8.15 years. Among the 15 patients (5 men), 13 had histological evidence of GCA. All the 15 patients had active disease at the time of blood sampling: twelve were included at the time of diagnosis, two experienced a disease relapse and another one had an acute flare while being treated with prednisone. None of the other 14 patients were taking corticosteroids at the time of blood sampling.

One-dimensional immunoblotting of IgG reactivity against VSMC protein extracts
One-dimensional immunoblots of IgG reactivity were analysed with VSMC protein extracts in sera from patients with GCA; control patients with ANCA-associated vasculitis, including those with WG, MPA and CSS; and HCs. All subjects tested expressed an IgG reactivity band directed against a 45-kDa protein. In patients with GCA, a number of IgG reactivities were expressed that were not identified in patients with ANCA-associated vasculitis or in HCs, including reactivities directed against protein bands of 85 kDa (Additional file 3).

Two-dimensional immunoblotting of IgG reactivity against VSMC protein extracts
The proteome of VSMCs contained 1,427 different proteins ranging from 3 to 10 isoelectrofocalisation points (IPs) and from 10 to 250 kDa. Among those, a mean (± SD) of 679 ± 258 protein spots were detected after being transferred onto polyvinylidene fluoride (PVDF) membranes. Serum IgG from the HC pool recognised 94 protein spots, whereas IgG from the 5 pools from GCA patients recognised a mean (± SD) of 100 ± 17 protein spots corresponding to a total of 268 different protein spots. Most of these 268 protein spots were recognised from only 1 or 2 pools from patients with GCA and/or from the HC pool. Among these protein spots, 29 were recognised by at least three-fifths of the pools from GCA patients, including 19 not recognised by the HC pool (Additional file 4, Supplemental Table  S2). These 19 protein spots were identified by MS as detailed in Table 1 and Additional file 5. The localisations of identified protein spots in the analytical gel are depicted in Figure 1. Among these proteins, only one, the far upstream element-binding protein 2 (FUBP2) (Figure 2), was recognised in all five pools of sera from  Table S2). Interestingly, IgG from pools of sera from each of three GCA patients recognised lamin A/C (Figure 3) and vinculin (Additional file 6).

IgG reactivity against HUVEC protein extracts
The proteome of HUVECs contains 820 different proteins ranging from 3 to 10 IP and from 10 to 250 kDa. Among these, a mean (± SD) of 515 ± 73 protein spots were successfully detected after transfer onto PVDF membranes. Serum IgG from the HC pool recognised 79 protein spots, whereas IgG from the 3 pools of GCA patients recognised a mean (± SD) of 162 ± 3 protein spots corresponding to 191 different protein spots. Most of these 191 protein spots were recognised in only 1 pool of IgG from GCA patients and/or were also recognised in the HC pool. Among these protein spots, 45 were recognised in at least two-thirds of pools from GCA patients, including 30 that were not recognised in the HC pool (Additional file 7, Supplemental Table S3). Of these 30 proteins, 22 were identified by matrixassisted laser desorption ionization time-of-flight/timeof-flight MS. Complete MS data are shown in Table 2.
Localisations of identified protein spots in the analytical gel are depicted in Figure 4. Overall, three proteins were recognised by IgG in sera from GCA patients in HUVEC and VSMC protein extracts: mitochondrial fumarate hydratase, lamin A/C and vinculin. IgG reactivity against vinculin and lamin A/C in sera from GCA patients and the HC pool are depicted in Figure 5 and Additional file 8 respectively.

Biological network analysis of identified autoantibody specificities
Lists of VSMC and HUVEC proteins specifically recognised and/or recognised with high intensity by IgG in sera from GCA patients were analysed with IPA. Interestingly, most of the VSMC and HUVEC proteins specifically recognised and/or recognised with high intensity interacted with growth factor receptor-bound protein 2 (Grb2), a protein involved in VSMC proliferation. Therefore, we could depict the signalling network between HUVEC and VSMC proteins identified as major targets of autoantibodies in patients with GCA ( Figure 6). Interestingly, TNF-α, IL-4 ( Figure 6) and other molecules such as platelet-derived growth factor and IFN-γ (Additional file 9) were also involved in this signalling network.

Discussion
In the present work, we detected IgG antibodies directed against the proteome of VSMCs and HUVECs in the sera of patients with GCA and identified their target antigens by using a 2-D immunoblotting technique and MS. Few studies have focused on perturbations of the humoral immune system in patients with GCA. Few B lymphocytes are detected in temporal artery biopsies from patients with GCA [30]. When present, they are mainly found in the adventitial layer [31]. Moreover, plasma cells can be found in the adventitia in 7% to 24% of temporal artery biopsies from patients with GCA [32]. Plasma cells might localise in adventitia because of an infectious agent initiating vascular inflammation. However, a number of studies failed to identify an  infectious agent, either a virus or bacteria, in the arterial wall by immunohistochemistry or PCR [33]. Alternatively, an autoantigen present in the arterial wall might trigger a specific immune response in GCA. AECAs have been detected in healthy individuals [34] and in a number of systemic autoimmune diseases [10,35]. AECAs have been associated with disease activity in patients with vasculitis, particularly in those with anti-ANCA-associated vasculitis, Takayasu's arteritis or GCA [15], although these data remain controversial [36]. In addition, AECA could induce EC apoptosis in patients with systemic sclerosis [37]. However, the pathogenic role of AECAs has not yet been documented in GCA, and further investigations are necessary in this clinical setting.
Although to our knowledge anti-VSMC antibodies have not yet been reported in a human disease, such antibodies have been identified in a mouse model of vasculitis. Baiu et al. [17] showed that splenic mouse lymphocytes cultured with syngenic VSMCs induced vasculitic lesions after adoptive transfer into these mice. Serum collected from mice with vasculitis contained antibodies directed against VSMCs. Both wild-type and B-cell-deficient mice showed vascular inflammation after serum transfer, but mice deficient in both B and T cells (Rag2 -/-) Yes it should did not, which suggests that immunoglobulin and cell-mediated pathways, particularly T cells, work in concert to contribute to the vasculitis lesions in this model. Thus, autoantibodies targeting proteins in the proteome of VSMCs might play a role in the pathogenesis of GCA, and their function needs to be further explored.
Few studies have been conducted to identify the potential targets of autoantibodies in GCA. Screening antigens in a cDNA library derived from normal human testis revealed high-intensity serum IgG reactivity  directed against a number of ubiquitous autoantigens, including human lamin C, cytokeratin and mitochondrial cytochrome oxidase subunit II in the sera of patients with GCA [38]. Interestingly, we identified vinculin, lamin A/C and mitochondrial fumarate hydratase as target antigens of antibodies to proteins in the proteome of VSMCs and HUVECs. Vinculin is a cytoskeleton protein involved in extracellular matrix adhesion and intercellular junctions by binding to actin filaments. This protein has several interaction sites with numerous  binding partners, including α-actin [39]. Changes in the relative content of vinculin and α-actin have been reported in the human aortic intima of patients with atherosclerosis [40]. Because vascular remodelling may occur in atherosclerosis, this type of change could be associated with vascular remodelling in GCA. Lamin A and C are both encoded by the LMNA gene and represent major constituents of the inner nuclear membrane. Mutations in the LMNA gene have been identified in a number of conditions, including Hutchinson-Gilford progeria syndrome [41]. The most frequent mutation responsible for progeria creates a truncated progeria mutant lamin A (progerin), which accumulates within the nuclei of human vascular cells and may be directly responsible for vascular involvement in progeria [42]. Other LMNA gene mutations, such as Dunnigan-type familial partial lipodystrophy (FPLD2), can lead to proatherogenic metabolic disturbances such as dyslipidemia, hyperinsulinemia, hypertension and diabetes. Premature atherosclerosis-induced FPLD2 seems to be associated with monogenic insulin resistance syndrome [43]. Identification of lamin A/C as target antigens in  sera from patients with GCA seems interesting. Further investigations are necessary to characterise the implications of lamin A/C in vascular remodelling in this condition. We found another antigen, far upstream elementbinding protein 2 (FUBP = KHSRP), to be recognised by all five pools of sera from GCA patients in VSMC protein extracts. This protein, which binds to a DNA region called far-upstream element (FUSE), is a transcriptional activator of c-myc, a proto-oncogene that plays a key role in the regulation of cell growth, proliferation and differentiation. The FUSE-binding protein (FUBP) regulates FUSE activity [44]. Anti-FUBP antibodies were identified in synoviocytes in patients with rheumatoid arthritis, a condition marked by a proliferation of synovial tissue [45]. However, the potential pathogenic role of these antibodies has not yet been identified.
By using IPA, we found that most of the VSMC and HUVEC proteins specifically recognised and/or recognised with high intensity by IgG in sera from GCA patients interacted with Grb2. Grb2 is an intracellular linker protein that facilitates the activation of the small GTPase Ras by receptor tyrosine kinases and is involved in VSMC proliferation. Zhang and colleagues [46] reported that Grb2 is required for the development of neointima in response to vascular injury. Thus, Grb2 might be overexpressed and/or activated in ECs and VSMCs of patients with GCA and might stimulate the remodelling process. Moreover, proteins overexpressed in the remodelling process in the presence of activated Grb2 might trigger a specific immune response, possibly through structural antigen modifications occurring in the presence of metalloproteases and/or reactive oxygen species. A number of the VSMC and HUVEC proteins specifically recognised and/or recognised with high intensity by IgG in sera from GCA patients interacted with TNF-α. This result is in agreement with the pathophysiology of GCA and the ongoing inflammatory process in the arterial wall.
The combined use of 2-DE and immunoblotting offers an interesting approach to the identification of target antigens of autoantibodies [25]. However, our work has several limitations. Fewer than 1,500 protein spots were stained in the reference gel of VSMC and HUVEC protein extracts, which is less than the total number of proteins contained in these cells. Therefore, a number of proteins were probably lost at each step of the technique, depending on their charge, molecular weight, subcellular localisation and/or abundance in the cell. In addition, as expected, none of the identified antigens represented cell surface proteins, because protein extraction for 2-DE does not permit identification of membrane proteins. Finally, our pools of sera were from three patients each because this number was sufficiently low to allow the detection of strong reactivity that would be present in the serum of a single individual [25]. However, we cannot rule out the possibility that a low-intensity reactivity specific to a given individual might not be detected by using this pooling approach.