Abstract
The objective of the present work was to set up a routine test adapted to screening for antiendothelial cell antibodies (AECAs) in serum samples with minimal interference from antinuclear antibodies (ANAs) or rheumatoid factors (RFs). We compared the titers of AECAs titrated following two enzyme-linked immunosorbent assays (ELISAs): (i) an ELISA with ethanol-fixed EA.hy926 monolayers as the antigenic substrate and (ii) an ELISA with nucleus-depleted lysates prepared from EA.hy926 cells and normalized for protein (1.0 to 1.7 mg/ml) and DNA (≤0.1 μg/ml) contents as a surrogate substrate (postnuclear supernatant ELISA [PNS-ELISA]). The AECA titers in 51 serum samples, including 28 samples containing ANAs, were compared. A significantly positive correlation (r = 0.77; P < 0.001) between the two series was shown only for the ANA-negative serum samples. Conversely, ANAs or RFs in samples were shown not to interfere in tests for AECAs by the PNS-ELISA. AECAs recognize their antigenic targets in postnuclear supernatants, which is representative of the endothelial antigenic content, with improvement of the reliability of the assay, a prerequisite to application of the assay for their evaluation in clinical practice.
Antiendothelial cell antibodies (AECAs), a heterogeneous group of autoantibodies, are associated with several diseases characterized by immune-mediated vascular damage (for reviews, see references 2 and 13), including systemic lupus erythematosus (19), systemic sclerosis (17), Wegener's granulomatosis (6), and rheumatoid arthritis complicated by vasculitis (7). Even if they recognize poorly characterized targets, they may be valuable as markers of disease activity, with a possible role in the pathophysiology of associated diseases, especially by inducing endothelial cell activation or apoptosis (2, 3, 12, 23). The presence of AECAs in the sera of some groups of patients may be an important etiopathogenic factor in the vasculopathies associated with the disorders mentioned above and classified as proposed by Praprotnik et al. (14). The association of AECAs with endothelial injury in the course of these diseases prompted us to develop assays for these antibodies in clinical practice, which in some cases requires histopathological examination of affected organs for confirmation.
Most assays usually used for the detection of AECAs involve human umbilical vein endothelial cells or endothelial cell lines seeded in microtiter plates for subsequent testing by enzyme-linked immunosorbent assay (ELISA) or assays in which AECAs are detected by immunofluorescence. Spurious increases in AECA titers may occur, e.g., due to anti-DNA autoantibodies, depending on an important cross-reactivity against endothelial cells (4); anti-heparan sulfate antibodies (16); or heterophile antibodies to the bovine serum proteins involved in the assay (18). One of the main problems in this field is the lack of agreement on a standardized method for detection of AECAs, with subsequent difficulty with interlaboratory comparisons (13, 24). In addition, in rheumatoid arthritis and Felty's syndrome, rheumatoid factor (RF) has been shown to increase nonspecific immunoglobulin binding to endothelial cells, with subsequent pitfalls in assays with RF-containing sera developed with endothelial cells (15).
In order to minimize these false-positive interferences and to propose a routine simple screening test for the detection of AECAs in patients with autoimmune vascular disorders, we developed a highly reproducible ELISA using a normalized cell lysate preparation. The detection of AECAs by this assay was found to be independent of the presence of antibodies with unrelated specificities, such as antinuclear antibodies (ANAs), antiactin antibodies, and RFs.
MATERIALS AND METHODS
Patient and control samples.
Serum samples were collected from patients with various non-organ-specific autoimmune disorders (including connective tissue disease, vasculitis, antiphospholipid syndrome, or viral infection) and were selected because their AECA concentrations covered a wide range, from 0.02 to 1.17 absorbance units (reference value, 0.2 absorbance units), as established by the cyto-ELISA described below. Sera from 40 healthy blood donors matched for age with the patients served as controls. All samples were kept frozen at −80°C until use.
Endothelial cells and preparation of cell lysates.
The human endothelial hybrid cell line EA.hy926 (a kind gift from C.-J. S. Edgell, University of North Carolina, Chapel Hill), obtained by fusing human umbilical vein endothelial cells with the human lung carcinoma cell line A549, was cultured in fetal calf serum (FCS) medium (Dulbecco modified Eagle medium supplemented with 2 mM glutamine, 1 mM pyruvate, 50 ml of hypoxanthine-aminopterin-thymidine per liter, and 10% [vol/vol] heat-inactivated FCS) at 37°C in a 5% CO2 atmosphere. Immediately after the cells in the cultures had reached confluence, the cells were detached by using a mixture of 0.25% (wt/vol) trypsin in 0.7 mM EDTA-150 mM NaCl and were washed once in FCS medium. All culture reagents were from Gibco (Cergy-Pontoise, France). The nonadherent EL-4 cells (a kind gift from J. L. Maryanski, Lausanne, Switzerland) were cultured in the same FCS medium.
The postnuclear supernatant (PNS) from cell lysates to be used as a source of autoantigens and to be developed in an ELISA was prepared as follows. The confluent cell monolayer was placed in a 25-cm2 flask (Falcon), washed three times with cold phosphate-buffered saline (PBS; pH 7.4) buffer, and submitted to lysis for 15 min on ice after the addition of 1 ml of cold lysis buffer (0.5% [wt/vol] Nonidet P-40, 0.1 M KCl, 0.01 M EDTA, 1 μM leupeptin, 1 μM pepstatin, 1 mM phenylmethylsulfonyl fluoride, 0.05 M Tris-HCl [pH 7.4]). The crude lysate was centrifuged in 1.5-ml Eppendorf tubes at 12,000 × g for 30 min at +4°C. The subsequent PNS was assayed for its protein content (MicroBCA; Pierce-Interchim, Montluçon, France) and DNA content (21). More than 1 mg of protein per ml was titrated from the PNS, and the PNS antigen preparation was thus suitable for use in the PNS-ELISA. It was stored at −80°C until use. All chemicals were of analytical grade.
ELISAs for AECAs.
Two assays were compared for the measurement of AECA titers: an already established cyto-ELISA, in which fixed endothelial cell monolayers constituted the target for antibodies, and a PNS-ELISA, developed with a normalized antigen preparation as a constant source of antigen. In both assays, the mean + 3 standard deviations (SDs) for 72 serum samples from healthy individuals was taken as the cutoff point of the test.
Ethanol-fixed EA.hy926 cells were used in the cyto-ELISA, as described previously (18). Briefly, EA.hy926 cells were plated (104 cells/well in FCS medium) in flat-bottom microtiter plates (Nunc, Kamstrup, Denmark). Confluent cell layers were washed with PBS buffer and fixed with 100 μl of absolute ethanol for 5 min at 4°C. After three washings in PBS buffer, the plates were blocked with PBS buffer containing 1% (wt/vol) bovine serum albumin and were then successively exposed to 100 μl of patient serum (diluted 1:100 in the same buffer) and to peroxidase-conjugated rabbit F(ab′)2 anti-human immunoglobulin G (IgG), IgM (diluted 1:4,000), and IgA (diluted 1:2,000) antibodies (Dako, Trappes France), followed by incubation with 0.2 g of o-phenylenediamine per liter, 0.5 ml of H2O2 per liter, and 0.05 M sodium phosphate (pH 5.0). For each serum sample, the mean absorbance at 492 nm (A492) for blank wells was subtracted from the mean values for the wells with EA.hy926 cells to account for nonspecific binding. Data were expressed in absorbance units (A492 for sample wells − A492 for negative control wells).
The PNS-ELISA was performed as follows. Each well of half of the microtiter plates (MaxiSorp; Nunc) was coated overnight at +4°C with 200 μl of the 10-μg/ml (protein content) PNS antigen preparation diluted in 0.1 M NaHCO3-Na2CO3 (pH 9.6) (antigen-coated wells), and each well of the other half of the plate was coated with 200 μl of buffer without antigen (blank wells). The plates were washed two times with the washing buffer (0.1% [wt/vol] Tween 20 in PBS buffer) and were exposed to the saturation buffer (1% [wt/vol] bovine serum albumin in washing buffer) for 60 min at room temperature. The serum samples were diluted 1:100 in the saturation buffer supplemented with 5 μg of PNS prepared from EL-4 cells per ml to prevent nonspecific binding and were preincubated for 15 min at room temperature. A total of 100 μl of each sample dilution was distributed in duplicate into antigen-coated and blank wells and incubated at 60 min at 30°C. Furthermore, we incorporated into each series a 1:100 dilution of a positive serum sample as a positive reference and a 1:100 dilution of a serum sample from a healthy donor as a negative control. After the plates were washed, 100 μl of peroxidase-conjugated goat F(ab′)2 anti-human IgG, IgA, and IgM (Cappel-Organon, Fresnes, France) diluted 1:12,000 in saturation buffer was added to each well; and the plates were incubated for 30 min at 30°C. After the plates were washed, 100 μl of ready-to-use tetramethylbenzidine substrate (Sigma, Chesnes, France) was added to each well and the plate was incubated for 15 min at 30°C. The reaction was stopped with 50 μl of 1 N H2SO4, and the A450 was measured. The mean A450 for blank wells was subtracted from the mean A450 for the antigen-coated wells to account for nonspecific binding.
Other autoantibodies.
ANAs and antiactin antibodies were detected and quantified by titration on HEp-2 cells (reference titer, 80; Euroimmun-BioAdvance, Emerainville, France), and extractable nuclear antigens (ENA) specificities were established by both ELISA (Euroimmun-BioAdvance or Pharmacia, St. Quentin-Yvelines, France) and dot blotting (Inolia-Ingen, Rungis, France). Anti-double-stranded DNA (anti-dsDNA) antibodies were titrated by ELISA (The Binding Site, Saint Egrève, France). Anti-human IgG RFs (IgM isotype; reference titer, 30) were measured as described by Gioud-Paquet et al. (5) by using an international standard (Bio-Rad Clinical Diagnostics, Marnes la Coquette, France).
Adsorption of AECA and ANA activities.
A total of 200 μl of serum (diluted 1:100 in saturation buffer) was incubated for 15 min on ice with either ethanol-fixed EA.hy926 cells (confluent cell layers were prepared as described above), 5 μg of bovine serum albumin, or 2 μg of PNS antigen. These incubations were repeated four times, and then the A450s were measured for determination of the AECA (PNS-ELISA) and anti-DNA antibody titers.
Statistical analysis.
We used conventional descriptive statistics to assess parameters of interassay variation and calculated Spearman's sampling correlation coefficient to compare two groups of quantitative data. When series of data from two assays were compared, they were considered positively correlated when rho (r) values were above 0.3 and strongly correlated when r values were above 0.7. P values of <0.05 were considered statistically significant.
RESULTS
Characterization of PNS preparations.
In order to develop a routine test for the detection of AECAs, we considered it to be of importance that the endothelial cell preparation involved as antigen in the assay be representative of the steady-state antigenic content of endothelial cells for all antigens except the nuclear and cytoskeletal antigens. This was achieved by a short incubation of EA.hy926 cells in the presence of mild detergent, keeping the integrity of the nuclei, according to technical recommendations for experimental development in cell biology (11). The nuclei and cytoskeletons were discarded in the pellets, and PNS was harvested as described in Material and Methods, which ensured that the antigen-normalized solution had protein concentrations ranging from 1.0 to 1.7 mg/ml and DNA concentrations less than 0.1 μg/ml.
Dose-response curves, cutoff assignment, and inter- and intra-assay variations.
The PNS prepared as described above was distributed in flat-bottom wells of microtiter plates in order to perform the PNS-ELISA, as described above in Material and Methods. To establish the working dilution of the serum samples, the dose-response curve for a positive control sample found to be positive by the cyto-ELISA, as described previously (18), exhibited a dose-dependent response in the PNS-ELISA, reaching a maximum A450 of 0.945 ± 0.035. A serum sample from a healthy donor found to be negative, as described above, had an A450 of 0.215 ± 0.018; the values for the blank wells for these positive and negative controls were 0.235 and 0.185, respectively. A representative example of the titration curves for the positive serum sample and the negative control serum sample from a healthy individual are shown in Fig. 1. The 1:100 dilution was used as the working dilution for all subsequent assays.
FIG. 1.
Dose-response curves for AECA-positive and -negative control sera. AECA titers were determined by the PNS-ELISA, as described in Materials and Methods.
To establish the cutoff value of the test, 72 serum samples from healthy donors were analyzed, and after subtraction of the values for the blank wells, A450 values ranged from 0.006 to 0.165 (mean + 1 SD, 0.070 ± 0.040). The cutoff value was calculated by taking the mean + 3 SDs, i.e., 0.190; this was assumed to give an index value of 100. As suggested by Rosenbaum et al. (19), the index was calculated by the equation 100 × [(S − A)/(C − A)], where S is the A450 value for the sample and A and C are the A450 values for the negative control and the control used to establish the cutoff value, respectively. Finally, the samples were classified as having negative (<100), medium-positive (100 to 250), and high-positive (>250) AECA titers.
The interassay variation was determined by calculating parameters for the descriptive statistics for the samples. Twenty serum samples were tested on different days, with AECA titers being above the reference value (medium to high titers) for 11 samples and below the reference value (low titers) for 9 samples. The corresponding coefficients of variation were from 1.1 to 18.9% for samples with medium to high AECA titers and from 8.2 to 22.6% for samples with low AECA titers (Table 1). The intra-assay variation was similarly determined with nine samples (four with medium to high titers and five with low titers), with coefficients of variation from 2.9 to 5.4% for samples with medium to high AECA titers and 3.2 to 8.9% for samples with low titers (data not shown).
TABLE 1.
Interassay variabilitylegend
| AECA titer group and sample no. | Mean titer | No. of assays | SD | Coefficient of variation (%) |
|---|---|---|---|---|
| Medium to high titer | ||||
| P1 | 151.6 | 6 | 18.4 | 12.2 |
| P2 | 132.0 | 2 | 8.4 | 6.4 |
| P3 | 117.0 | 3 | 22.0 | 18.9 |
| P4 | 139.5 | 2 | 19.0 | 13.7 |
| P5 | 198.0 | 2 | 17.0 | 12.1 |
| P6 | 324.0 | 2 | 29.6 | 9.2 |
| P7 | 255.0 | 3 | 11.3 | 4.4 |
| P8 | 192.5 | 2 | 2.1 | 1.1 |
| P9 | 238.7 | 3 | 32.8 | 13.8 |
| P10 | 125.4 | 5 | 19.4 | 15.5 |
| P11 | ||||
| Low titer | ||||
| N1 | 40.2 | 5 | 6.9 | 17.4 |
| N2 | 36.4 | 5 | 8.2 | 22.6 |
| N3 | 45.8 | 5 | 8.0 | 17.5 |
| N4 | 57.5 | 2 | 4.9 | 8.6 |
| N5 | 24.3 | 3 | 4.7 | 19.4 |
| N6 | 89.5 | 2 | 7.7 | 8.7 |
| N7 | 85.7 | 3 | 7.0 | 8.2 |
| N8 | 24.0 | 4 | 2.1 | 17.9 |
| N9 | 24.2 | 4 | 4.3 | 17.9 |
aDescriptive statistics were determined with data for 20 samples representative of those with medium to high AECA titers (titers, >100; n = 11 samples [samples P1 to P11]) and low AECA titers (titers, <100; n = 9 samples [samples N1 to N9]) tested two to six times on different days by the PNS-ELISA.
Comparison of AECA titers by cyto-ELISA and PNS-ELISA. (i) Interference of ANA.
Fifty-one serum samples submitted for routine testing for AECAs, including 28 serum samples that were ANA positive (ANA titers, >160) and that were chosen because of their ranges of AECA titers, were assayed by both the cyto-ELISA and the PNS-ELISA. The ANA-positive sera expressed antibodies of diverse specificities (PM-Scl, RNP, U1-RNP, SS-A, SS-B, centromere, PCNA, dsDNA). We first compared the AECA titers in all 51 samples obtained by both assays and found a correlation coefficient (r) of 0.33 (P < 0.02) (Fig. 2), indicating a weak correlation between the two methods. We then compared both ELISAs for their abilities to detect AECAs in the 23 ANA-negative samples and observed an r value of 0.77 (P < 0.001) (Fig. 2), indicating in this case a good correlation of the data between the cyto-ELISA and the PNS-ELISA.
FIG. 2.
Correlation between cyto-ELISA and PNS-ELISA. AECA titers in 51 serum samples containing (♦) or not containing (⋄) ANAs at significant titers (titers, ≥160; reference titer, 80) were assayed. The ELISAs were performed with fixed EA.hy926 cell monolayers (cyto-ELISA, horizontal axis; reference value, 0.2 absorbance [A] unit) and a PNS antigen preparation (PNS-ELISA, vertical axis; reference value, 100). According to Spearman's rank test for comparison of the two series of data, r values were 0.33 (P < 0.02) for all data and 0.77 (P < 0.001) for data obtained from samples negative for ANAs (open diamonds).
The 28 ANA-positive serum samples appeared to be positive by the cyto-ELISA. A comparison of the AECA and ANA titers led to an r value of 0.65 (P < 0.001) when the values from the cyto-ELISA were used and an r value of 0.20 (P = 0.310) when the values from the PNS-ELISA were used. This suggests an independence between AECA and ANA titers when PNS is used as a substrate but not when fixed EA.hy926 cells are used. When data for all ANA-positive samples were excluded, 16 samples were found to be positive by the PNS-ELISA, whereas 14 were found to be positive by the cyto-ELISA, indicating that the capacities to discriminate between negative and positive samples remained similar for both assays.
We next analyzed data for 50 serum samples with ranges of AECA titers (evaluated by PNS-ELISA) and ANA titers. We observed a nonsignificant r value of 0.13 between the AECA and the ANA titers (P = 0.38) (Table 2), indicating the absence of a correlation between the AECA (PNS-ELISA) and ANA titers.
TABLE 2.
Interference between RF, ANA, and AECA titersa
| Patient no. | Patient disease | Titer
|
ENA specificityc | ||
|---|---|---|---|---|---|
| AECA (index) | RF (IU/ml) | ANAb | |||
| 1 | Phlebitis | 72 | 120 | <80 | |
| 2 | Prostate resection | 68 | 125 | <80 | |
| 3 | Fibromyalgia | 155 | 15 | <80 | |
| 4 | Renal transplantation | 51 | 80 | <80 | |
| 5 | Churg-Strauss syndrome | 90 | 1,200 | <80 | |
| 6 | Churg-Strauss syndrome | 37 | 150 | <80 | |
| 7 | Lung embolism | 92 | 70 | <80 | |
| 8 | Sjögren's syndrome | 646 | 290 | <80 | |
| 9 | Sjögren's syndrome | 25 | 200 | <80 | |
| 10 | Rheumatoid arthritis | 22 | 270 | <80 | |
| 11 | Rheumatoid arthritis | 82 | 125 | <80 | |
| 12 | Rheumatoid arthritis | 52 | 1,750 | <80 | |
| 13 | Rheumatoid arthritis | 66 | 410 | <80 | |
| 14 | Rheumatoid arthritis | 34 | 90 | <80 | |
| 15 | 118 | 260 | <80 | ||
| 16 | Rheumatoid arthritis | 55 | 100 | <80 | |
| 17 | Rheumatoid arthritis | 17 | 115 | <80 | |
| 18 | Rheumatoid arthritis | 33 | 240 | <80 | |
| 19 | Rheumatoid arthritis | 25 | 330 | <80 | |
| 20 | Rheumatoid arthritis | 91 | 165 | <80 | |
| 21 | Rheumatoid arthritis | 202 | 1,500 | <80 | |
| 22 | Rheumatoid arthritis | 231 | 380 | <80 | |
| 23 | Rheumatoid arthritis | 118 | 110 | <80 | |
| 24 | Rheumatoid arthritis | 176 | 195 | <80 | |
| 25 | Rheumatoid arthritis | 78 | 1000 | 320 (h) | 0 |
| 26 | Rheumatoid arthritis | 33 | 5,600 | 160 (h, nue) | 0 |
| 27 | Sjögren's syndrome | 39 | 140 | 320 (speck) | SS-A |
| 28 | Sjögren's syndrome | 23 | 170 | >1,280 (speck) | SS-A |
| 29 | Sjögren's syndrome | 59 | 500 | >1,280 (speck) | SS-A/SS-B |
| 30 | Sjögren's syndrome | 90 | 740 | >1,280 (speck) | SS-A |
| 31 | Systemic sclerosis | 80 | 385 | >1,280 (centro) | CENP-B |
| 32 | Systemic sclerosis | 79 | 350 | >1,280 (h, nuc) | PM-Scl |
| 33 | Systemic sclerosis | 41 | 765 | >1,280 (h) | U1-RNP |
| 34 | Systemic sclerosis | 77 | 300 | >1,280 (centro) | CENP-B |
| 35 | Systemic sclerosis | 144 | 100 | >1,280 (centro) | CENP-B |
| 36 | Systemic sclerosis | 26 | 40 | >1,280 (centro) | CENP-B |
| 37 | Severe Raynaud's phenomenon | 113 | 765 | >1,280 (h) | U1-RNP |
| 38 | Sclerosis polymyositis overlap syndrome | 32 | 90 | >1,280 (nuc) | PM-Scl |
| 39 | Lupus | 34 | 250 | >1,280 (speck) | SS-A |
| 40 | Lupus | 189 | 95 | >1,280 (speck) | PCNA |
| 41 | Lupus | 257 | 40 | 160 (h) | 0 |
| 42 | Lupus | 158 | 35 | 640 (speck) | SS-A |
| 43 | Lupus | 165 | 25 | >1,280 (h) | dsDNA (200 IU/ml) |
| 44 | Mixed connective tissue disease | 56 | 560 | 640 (speck) | 0 |
| 45 | Mixed connective tissue disease | 113 | 45 | >1,280 (speck) | RNP |
| 46 | Mixed connective tissue disease | 252 | 40 | >1,280 (speck) | SS-A/RNP |
| 47 | Primary antiphospholipid syndrome | 199 | 95 | >1,280 (h) | dsDNA (850 IU/ml) |
| 48 | Primary antiphospholipid syndrome | 153 | 30 | 160 (centro) | CENP-B |
| 49 | Primary antiphospholipid syndrome | 305 | 25 | 160 (h) | 0 |
| 50 | Hashimoto's thyroiditis | 27 | 160 | 160 (h) | 0 |
The AECA (PNS-ELISA) anti-human IgG RF, and ANA titers in 50 serum samples were assessed by the assays described in Materials and Methods. Comparisons of the data obtained by the assays for AECA and ANA and the assays for AECA and RF were performed according to Spearman's rank test; r values were 0.13 (P = 0.38) and −0.20 (P = 0.17), respectively. The reference values for AECA, RF, and ANA are <100, <30 IU/ml, and <80, respectively.
HEp-2 patterns of ANA fluorescence are indicated as follows: centro, centromere; h, homogeneous; nuc, nucleolar; speck, speckled.
0, none.
The binding of ANA to the PNS antigen was compared with that to the fixed EA.hy926 cells. An absorption experiment with an ANA-positive sample with anti-DNA specificity was carried out, and the resulting AECA and anti-DNA reactivities were determined. After repeated incubations with fixed EA.hy926 cells, the binding of the antibodies from the sample to the cell antigen resulted in a decrease in the anti-DNA-related A450 that was more than that for the PNS antigen (∼30% difference), suggesting that EA.hy926 cells partially removed the ANA activity from the patient's sample (Fig. 3). Taken together, these results strongly suggest that ANA should not interfere in the PNS-ELISA for AECAs.
FIG. 3.
Absorption of AECA and ANA activities from a patient serum sample by fixed EA.hy926 cells and PNS antigens. A sample from patient 47 (Table 2) diluted 1:100 was submitted to zero to four repeated incubations in the presence of fixed EA.hy926 cells (A), PNS antigen (B), and bovine serum albumin as a control (open symbols in both panels). The ELISAs were performed with PNS antigen (squares) and dsDNA (triangles).
(ii) Interference of RFs and antiactin antibodies with AECA titers.
RFs are likely to give false-positive results when serum is analyzed by ELISA, in particular, in assays for AECAs with endothelial cells (15). In order to be eligible for use with samples from patients with autoimmune diseases in which RFs are present, the PNS-ELISA for AECAs was developed with the 50 samples containing IgM RFs (up to 5,600 IU/ml) described above. The data are summarized in Table 2. Comparison of the AECA and RF titers gave r values of −0.20 (P = 0.17), indicating an absence of a correlation between the two titers, with a total independence of the AECA and RF titers found in the sera. This result indicates that RFs do not interfere in the PNS-ELISA described for AECAs in this study.
Antiactin-positive sera scored negative for AECAs, suggesting that antiactin antibodies do not interfere in the test (data not shown).
DISCUSSION
A number of methods have been developed for the detection of AECAs, including indirect immunofluorescence, radioimmunoassay, Western blotting, assays with cell membranes, and cyto-ELISA. The cell substrates have included endothelial cells derived from different origins, namely, arteries, saphenous or umbilical veins, and endothelial cell lines.
It became important to overcome variations in differences in the tests owing to differences in the endothelial cells (donor variability or cell number), with subsequent high coefficients of variation between the assays. This prompted us to develop an alternative to the cyto-ELISA with an antigenic target of AECAs in a technique that can be used for routine testing after normalization of the antigenic preparation for protein and DNA contents. The advantage of the present ELISA is the fact that the PNS antigen from EA.hy926 cells is a constant and easily accessible source of antigen. Comparison of the results obtained by the two ELISAs, the cyto-ELISA and the PNS-ELISA, shows that they have a good correlation when ANA-negative sera are used, as assessed by Spearman's sampling correlation coefficient. This is in contradiction to the data reported by Heurkens et al. (8), who used sonicates from cell membranes prepared after scraping of culture monolayers. In that assay the poor correlation observed between the cyto-ELISA and the cell membrane ELISA was assumed to be due to the loss or denaturation of surface antigens. The surrogate preparation used in the present study is representative of the endothelial cell content, including already observed antigenic targets of AECAs expressed as membrane-associated proteins at steady state, as demonstrated by adsorption experiments (8), or expressed within the cytoplasms of endothelial cells (6). Recent experiments attempted to characterize the antigenic targets of AECA in relation to one disease or another; it was concluded that the targets were a mosaic of constitutive antigenic determinants (for a review, see reference 14), for example, the cytoskeletal proteins tubulin and vimentin and the membrane-associated protease collagenase identified by immunobloting (22), CD36 and a 50-kDa keratin-like protein identified after immunoscreening of a serum sample with a cDNA expression library (1), and an unidentified 70-kDa structure isolated after immunoprecipitation of iodinated membrane protein targets (9). Therefore, it was of importance to consider the PNS as a whole antigenic target and, consequently, to prevent during its preparation the loss of any cell proteins or protein associations by the addition of protease inhibitors in the lysis buffer and by the use of a low temperature during each preparation step.
When we used the cyto-ELISA with fixed endothelial cells, we incidentally noticed high-positive ELISA responses with ANA-positive sera, suggesting that the antibody detected could be related to nuclear antigens and, hence, nonspecific for endothelial cells. This is of particular interest when samples from patients with systemic sclerosis or lupus erythematosus are assayed for AECAs. Indeed, in patients with such systemic disorders, ANAs are frequently found at high titers and AECAs may contribute or may be associated with the pathogenesis and the activity of the disease (3, 10, 20). The experimental procedure that we retained to circumvent the problem of false-positive results due to ANAs and to properly detect associated genuine AECAs was to discard the nucleus from the PNS antigen preparation. The number of samples which were positive for both AECAs and ANAs was subsequently reduced (Fig. 2), and AECA and ANA titers did not correlate (Table 2). To rule out possible nonspecific reactivity because of intracellular lectins or anticytoplasm antibodies, prior to the ELISA the sample was absorbed in diluent containing PNS prepared from EL4 cells. Moreover, the test result was found to be independent of the presence of RF in the sample, a frequent origin of false-positive results by assays for autoantibodies.
The frequency of detection of AECAs depends on the method developed for their detection. One of the main problems in this field is the lack of agreement on a standardized antigen, with subsequent difficulty with interlaboratory comparisons. The preparation of a PNS from the EA.hy926 cell line normalized for protein and DNA contents could allow better interlaboratory agreement in the measurement of AECA titers by ELISA and allow the data obtained for the same sera processed by different assays to be compared, as recommended by Youinou et al. (24). In addition, this could contribute to a reduction in the important discrepancies in associations between AECA titers and clinical conditions ascribed to antigen variability (13, 2, 9). Better knowledge of the prevalence of AECAs in various autoimmune diseases and improved interpretation of the data could take advantage of this standardization of the endothelial cell antigen and its subsequent use in a routine ELISA.
Acknowledgments
This work was supported by grant 1452 from the Délégation Régionale à la Recherche Clinique 1999.
We are indebted to G. Dubos, C. Pison, J.-P. Zarski, and J.-P. Brion (Grenoble, France) for providing us with serum samples and to J.-Y. Cesbron for continuous encouragement.
REFERENCES
- 1.Ationu, A. 1998. Identification of endothelial antigens relevant to transplant coronary artery disease from a human endothelial cell cDNA expression library. Int. J. Mol. Med. 1:1007-1010. [DOI] [PubMed] [Google Scholar]
- 2.Belizna, C., and J. W. Cohen Tervaert. 1997. Specificity, pathogenicity, and clinical value of endothelial cell antibodies. Semin. Arthritis Rheum. 27:98-109. [DOI] [PubMed] [Google Scholar]
- 3.Bordron, A., M. Dueymes, Y. Levy, C. Jamin, J.-P. Leroy, J.-C. Piette, Y. Shoenfeld, and P. Y. Youinou. 1998. The binding of some human anti-endothelial cell antibodies induces endothelial cell apoptosis. J. Clin. Investig. 101:2029-2035. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Chan, T. M., and K. P. Chen. 1997. Identification of endothelial cell membrane proteins that bind anti-DNA antibodies from patients with systemic lupus erythematosus by direct or indirect mechanisms. J. Autoimmun. 10:433-439. [DOI] [PubMed] [Google Scholar]
- 5.Gioud-Paquet, M., M. Auvinet, T. Raffin, P. Girard, M. Bouvier, E. Lejeune, and J.-C. Monier. 1987. IgM rheumatoid factor (RF), IgA RF, IgE RF, and IgG RF detected by ELISA in rheumatoid arthritis. Ann. Rheum. Dis. 46:65-71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Gobel, U., J. Eichhorn, R. Kettritz, L. Briedigkeit, D. Sima, C. Lindschau, H. Haller, and F. C. Luft. 1996. Disease activity and autoantibodies to endothelial cells in patients with Wegener's granulomatosis. Am. J. Kidney Dis. 28:186-194. [DOI] [PubMed] [Google Scholar]
- 7.Heurkens, A. H. M., P. S. Hiemstra, G. J. M. Lafeber, M. R. Daha, and F. C. Breedveld. 1989. Anti-endothelial antibodies in patients with rheumatoid arthritis complicated by vasculitis. Clin. Exp. Immunol. 78:7-12. [PMC free article] [PubMed] [Google Scholar]
- 8.Heurkens, A. H. M., A. Gorter, T. M. de Vreede, C. S. Edgell, F. C. Breedveld, and M. R. Daha. 1991. Methods for the detection of anti-endothelial antibodies by enzyme-linked immunosorbent assay. J. Immunol. Methods 141:33-39. [DOI] [PubMed] [Google Scholar]
- 9.Levy, Y., B. Gilburd, J. George, N. Del Papa, R. Mallone, M. Damianovich, M. Blank, A. Radice, Y. Renaudineau, P. Youinou, A. Wiik, F. Malavasi, P.-L. Meroni, and Y. Shoenfeld. 1998. Characterization of murine monoclonal anti-endothelial cell antibodies (AECA) produced by idiotypic manipulation with human AECA. Int. Immunol. 10:861-868. [DOI] [PubMed] [Google Scholar]
- 10.Li, J. S., M. F. Liu, and H. Y. Lei. 1996. Characterization of antiendothelial cell antibodies in the patients with systemic lupus erythematosus: a potential marker for disease activity. Clin. Immunol. Immunopathol. 79:211-216. [DOI] [PubMed] [Google Scholar]
- 11.Lukas, J., and J. Bartek. 1998. Immunoprecipitation of proteins under nondenaturing conditions, p. 489-494. In J. E. Celis (ed.), Cell biology: a laboratory handbook, vol. 4, 2nd ed. Academic Press, London, United Kingdom.
- 12.Lunardi, C., C. Bason, R. Navone, E. Millo, G. Diamonte, R. Corrocher, and A. Puccetti. 2000. Systemic sclerosis immunoglobulin G autoantibodies bind the human cytomegalovirus late protein UL94 and induce apoptosis in human endothelial cells. Nat. Med. 6:1183-1186. [DOI] [PubMed] [Google Scholar]
- 13.Meroni P.-L., M.-A. Khamashta, P. Youinou, and Y. Schoenfeld. 1995. Mosaic of anti-endothelial antibodies. Review of the first international workshop on anti-endothelial antibodies: clinical and pathological significance. Lupus 4:95-99. [DOI] [PubMed] [Google Scholar]
- 14.Praprotnik, S., M. Blank, P.-L. Meroni, B. Rozman, A. Eldor, and Y. Shoenfeld. 2001. Classification of anti-endothelial cell antibodies into antibodies against microvascular and macrovascular endothelial cells: the pathogenic and diagnostic implications. Arthritis Rheum. 44:1484-1494. [DOI] [PubMed] [Google Scholar]
- 15.Quadros, N. P., P. J. Roberts-Thomson, and A. S. Gallus. 1990. IgG and IgM anti-endothelial cell antibodies in patients with collagen-vascular disorders. Rheumatol. Int. 10:113-119. [DOI] [PubMed] [Google Scholar]
- 16.Renaudineau, Y., R. Révélen, A. Bordron, D. Mottier, P. Youinou, and R. Le Corre. 1998. Two populations of endothelial cell antibodies cross-react with heparin. Lupus 7:86-94. [DOI] [PubMed] [Google Scholar]
- 17.Renaudineau, Y., R. Révélen, Y. Levy, K. Salojin, B. Gilburg, Y. Shoenfeld, and P. Youinou. 1999. Anti-endothelial cell antibodies in systemic sclerosis. Clin. Diagn. Lab. Immunol. 6:156-160. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Révélen, R., A. Bordron, M. Dueymes, P. Youinou, and J. Arvieux. 2000. False positivity in a cyto-ELISA for anti-endothelial cell antibodies caused by heterophile antibodies to bovine serum proteins. Clin. Chem. 46:273-278. [PubMed] [Google Scholar]
- 19.Rosenbaum, J., B. E. Pottinger, P. Woo, C. M. Black, S. Loizou, M. A. Byron, and J. D. Pearson. 1988. Measurements and characterization of circulating anti-endothelial cell IgG in connective tissue diseases. Clin. Exp. Immunol. 72:450-456. [PMC free article] [PubMed] [Google Scholar]
- 20.Salojin, K. V., M. Le Tonqueze, A. Saraux, E. L. Nassonov, M. Dueymes, J.-C. Piette, and P. Y. Youinou. 1997. Antiendothelial cell antibodies: useful markers of systemic sclerosis. Am. J. Med. 102:178-185. [DOI] [PubMed] [Google Scholar]
- 21.West, D. C., A. Sattar, and S. Kumar. 1985. A simplified in situ solubilization procedure for the determination of DNA and cell number in tissue cultured mammalian cells. Anal. Biochem. 147:289-295. [DOI] [PubMed] [Google Scholar]
- 22.Yazici, Z. A., M. Behrendt, D. Cooper, M. Goodfield, L. Partridge, and N. J. Lindsey. 2000. The identification of endothelial cell autoantigens. J. Autoimmun. 15:41-49. [DOI] [PubMed] [Google Scholar]
- 23.Yazici, Z. A., E. Raschi, A. Patel, C. Testoni, M. O. Borghi, A. M. Graham, P.-L. Meroni, and N. Lindsey. 2001. Human monoclonal anti-endothelial cell IgG-derived from a systemic lupus erythematosus patient binds and activates human endothelium in vitro. Int. Immunol. 13:349-357. [DOI] [PubMed] [Google Scholar]
- 24.Youinou, P., P.-L. Meroni, M. A. Khamashta, and Y. Schoenfeld. 1995. A need for standardization of the anti-endothelial cell antibody test. Immunol. Today 16:363-364. [DOI] [PubMed] [Google Scholar]