To determine whether patients with rheumatoid arthritis (RA) have elevated Epstein-Barr virus (EBV) load in their peripheral blood mononuclear cells (PBMCs) and whether it is correlated with the HLA–DR genes they express, we developed an accurate EBV DNA quantitative assay using real-time polymerase chain reaction (PCR) with fluorescent probes.
We studied the EBV DNA load in the PBMCs of 84 patients with RA, 69 normal controls, and 22 patients with rheumatic conditions other than RA. A 214-bp segment from the long internal repeat of EBV was amplified from 500 ng of PBMC DNA (150,000 cells) and quantified by real-time PCR with fluorescent probes.
We demonstrated that in patients with RA, the EBV DNA load in PBMCs is increased almost 10-fold compared with that in normal controls. The EBV load is stable over time and is not obviously influenced by disease-modifying antirheumatic drugs or HLA–DR.
Patients with RA have elevated EBV load in their peripheral blood.
In patients with rheumatoid arthritis (RA), control of Epstein-Barr virus (EBV) infection is slightly impaired. Indeed, sera from patients with RA contain high-titer antibodies to latent or replicative EBV antigens (1). T lymphocytes from the peripheral blood of patients with RA are not efficient at controlling the outgrowth of EBV-positive B lymphoblastoid lines (2). Finally, the frequency of EBV-infected peripheral blood B lymphocytes is higher in patients with RA than in controls (3).
RA develops on the HLA–DR4/DR1 background, and it has been suggested that this background might also be responsible for deficient control of EBV infection (4). To test this hypothesis, we previously calculated the frequencies of peripheral blood T lymphocytes specific for a major replicative EBV antigen, gp110 (5). We observed that HLA–DRB1*0404, an allele predisposing to RA, was associated with low frequencies of gp110-specific T cell precursors. In contrast, HLA–DRB*07, a protective allele, was associated with the highest frequencies of gp110-specific precursors (5). This suggested that the HLA–DR background predisposing to RA could be responsible for poor T cell responses to EBV gp110, potentially leading to EBV overload. Still, even if poor control of EBV infection in patients with RA has been repeatedly observed, in the absence of a really quantitative EBV DNA assay, EBV overload in these patients has never been properly demonstrated and quantified.
Here, we used real-time quantitative polymerase chain reaction (PCR) to develop an accurate assay of EBV DNA and to evaluate EBV load in peripheral blood mononuclear cells (PBMCs). We observed that EBV load is increased in patients with RA and that it is not correlated with any particular HLA background.
PATIENTS AND METHODS
Patients and controls.
We studied 135 peripheral blood samples obtained from 84 patients satisfying the 1987 American College of Rheumatology (formerly, the American Rheumatism Association) criteria for RA (6). Forty patients were treated with infliximab, 3 were treated with etanercept, 33 were treated with methotrexate, 9 were treated with miscellaneous disease-modifying antirheumatic drugs (including sulfasalazine, gold salts, and hydroxychloroquine), and 14 patients received no treatment. Fourteen patients received more than 1 treatment or changed treatments during the study. Sixty-nine healthy controls with no history of RA were chosen from bone marrow donors at the Marseille Etablissement Français du Sang and from the staff of the immunorheumatology laboratory. Twenty-two patients with inflammatory rheumatic diseases different from RA were also tested (5 patients with ankylosing spondylitis, 6 with psoriatic arthritis, 2 with Behçet's disease, 2 with Sjögren's syndrome, 2 with inflammatory arthritis after BCG therapy, 2 with undifferentiated arthritis, 1 with polymyalgia rheumatica, 1 with periodic fever, and 1 patient with ulcerative colitis and arthritis).
For every patient and control, HLA–DR typing was performed by PCR/sequence-specific oligonucleotide analysis. To look for variation in EBV load, we obtained serial samples from 10 patients and 5 controls, within a period of 1 year or longer. Finally, to compare EBV load between HLA–DRB1*0404 and HLA–DRB1*07 positive patients, we added 19 new patients (7 expressing DRB1*0404, 12 expressing DRB1*07) to our starting series. This addition allowed comparison of EBV load in 18 RA patients expressing DRB1*0404 and 21 RA patients expressing DRB1*07.
Human genomic DNA was isolated from 5–10 ml of heparinized blood. Mononuclear cells were isolated by isopyknic centrifugation through Ficoll-Histopaque (Sigma, St. Louis, MO) and processed through Qiagen Genomic-tips 100/G (Qiagen, Courtaboeuf, France) according to the Qiagen genomic DNA handbook. DNA was resuspended in 10 mM Tris, pH 8, and was quantified by real-time PCR using a LightCycler (Roche Diagnostics, Mannheim, Germany), as previously described (7).
Quantification of EBV copy number.
A 214-bp segment from the highly conserved long internal repeat region 1 (IR-1) of EBV was amplified by real-time PCR. The Raji cell line, which harbors 50 copies of EBV genome per cell, was used as an external EBV standard. EBV copy number was calibrated by serial dilutions of Raji cell DNA, and ranged from 105 to 0.064 copies per microliter of standard EBV DNA dilution. Standard dilutions were stored at 4°C. Because 4 μl of standard DNA was used in experiments, external EBV standards contained 4 × 105 to 0.26 copies of EBV. Because IR-1 is repeated 10 times on the EBV genome, 4 μl of standard DNA contained 40 × 105 to 2.6 copies of the internal repeat target sequence.
Preliminary studies had shown that dilution of Raji DNA in EBV-free genomic DNA rather than distilled water is not necessary for quantification of EBV in 500 ng of genomic DNA (results not shown). Approximately 500 ng (from 1.5 × 105 lymphocytes) of test DNA obtained from patients or controls was used for EBV quantitation. To reduce pipetting errors, at least 4 μl of test DNA was used in all experiments. Each sample was tested in duplicate, and quantitation was performed twice. Optimal amplification conditions were established with SYBR Green I technique (data not shown), and specific quantification was performed with hybridization probes. The following oligonucleotides and probes were designed and synthesized by TIB Molbiol (Berlin, Germany): EBV forward (CCG-AAA-TCT-GAC-ACT-TTA-GAG-C), EBV reverse (CCC-TGA-CCT-TTG-GTG-AAG-TC), EBV fluorescein probe (GCC-TAA-AAC-CCC-CAG-GAA-GCG-G X), and EBV red 640 probe (TCT-ATG-GTT-GGC-TGC-GCT-GCTG p). PCR mix was prepared with the FastStart hybridization kit (Roche Molecular Biochemicals, Mannheim, Germany). The 20-μl final reaction volume contained 3 mM of MgCl2, 0.5 μM of each primer, 3 pM of LC red 640 probe, 3 pM of fluorescein probe, and 1 unit of heat-labile uracil-DNA-glycosylase (Roche Molecular Biochemicals). PCR amplification was performed as follows: 10 minutes of denaturation at 95°C followed by 5 cycles of 10 seconds at 98°C, 10 seconds at 56°C, 10 seconds at 72°C, and 40 cycles of 10 seconds at 95°C, 10 seconds at 56°C, 10 seconds at 72°C. Fluorescence signal was measured at the end of the annealing step.
Fluorescence data appear as a curve, plotting fluorescence against the number of cycles of PCR. From these data, concentration is calculated with LightCycler software using the fit-points analysis method. In short, 2 points from the log-linear part of the fluorescence curve are selected to define a log-linear fluorescence line, whose intersection with the background line is called a crossing point (indicating the number of cycles necessary to get a log-linear fluorescence curve). A standard curve is obtained by plotting crossing points for known concentrations of standard DNA. The crossing point of each sample is then used to calculate a concentration by using the standard curve (8).
Comparisons between groups were performed by Student's t-test and Kruskal-Wallis tests, using SAS software (SAS Institute, Cary, NC). To test for correlation between the 28-joint Disease Activity Score (DAS28) (9) and EBV load, we plotted DAS28 scores against EBV loads in 20 patients and calculated the Pearson's product-moment correlation coefficient for these data (10).
Data obtained using hybridization probes were analyzed by the fit-points method (8). The standard curve was linear over at least 4 orders of magnitude (Figure 1). All quantifications of EBV load were analyzed after importing this standard curve, thus providing comparable values for all experiments. The standard dilution containing 1.28 copies of EBV genome (13 copies of internal repeat target sequence) always gave a fluorescent signal; in the dilution containing 0.26 copy of EBV genome (2.6 copies of internal repeat target sequence), it was inconsistently detected. For test samples, we considered that a result as low as 0.1 copy of EBV genome was acceptable, because it corresponded to 1 copy of the internal repeat target sequence. For final results, EBV copy numbers in test samples were calculated for 500 ng of DNA (1.5 × 105 cells). It should be noted that some variation may exist in the number of copies of the first internal repeat on the EBV genome between different EBV isolates: a majority of isolates contain 10 repeats, but some rare isolates may have as few as 6 repeats (11, 12). Thus, using IR-1 to detect EBV DNA improves sensitivity but could, in rare cases, cause <2-fold underevaluation of EBV load.
All samples had been tested previously for DNA quantification by real-time PCR. Thus, lack of amplification was not attributable to PCR inhibitors. The specificity of primers and probes was tested in DNA obtained from healthy, EBV-negative donors, in EBV-negative Jurkat cell DNA (DSMZ ACC 282; American Type Culture Collection, Rockville, MD), and in DNA from purified EBV (data not shown).
Higher EBV load in patients with RA than in controls.
We could detect EBV in 88% of patients with RA, 93% of patients with inflammatory rheumatic conditions other than RA (non-RA patients), and 89% of healthy controls. In most patients and a few controls, the copy number is an average value of multiple quantifications performed during followup. EBV loads for 1.5 × 105 PBMCs range from 0 to 185 copies in patients with RA, from 0 to 43.7 copies in non-RA patients, and from 0 to 34 copies in healthy controls. In patients with RA, the mean EBV copy number was 15.64 copies per 150,000 PBMCs (median 8.84 copies). In non-RA patients, the mean EBV copy number was 5.73 copies per 150,000 PBMCs (median 1.56 copies). In healthy controls, the mean EBV copy number was 1.89 (median 0.6 copy) (Figure 2). EBV load was significantly different between RA patients and controls (P = 0.0002) and between RA patients and non-RA patients (P = 0.0045), but was not significantly different between non-RA patients and normal controls (P = 0.057).
EBV load followup in patients and controls.
Influence of disease activity on EBV load in RA patients.
To test whether disease activity influences EBV load in patients with RA, we plotted 37 DAS28 scores of 23 patients, obtained at different times, against the EBV load in their peripheral blood at the same times. No correlation between the two was observed (R2 = 0.08) (data not shown).
Influence of treatment on EBV load in RA patients.
No significant difference in EBV load was associated with use of methotrexate, methotrexate plus infliximab, or intramuscular gold salts. We looked for remarkable variation of EBV load in every patient. Followup of >1 year in 11 RA patients receiving methotrexate revealed no significant variation of EBV load (Figure 3). A zigzag pattern or a flat curve was usually seen. In most patients, the combination of infliximab and methotrexate did not seem to influence EBV load either, but the followup time was shorter in this group of patients (Figure 4).
Follow-up in controls.
Five healthy controls were followed up for a 1-year period. No variation in EBV load was observed (data not shown).
Factors influencing EBV load in patients with RA.
Age at the time of diagnosis, duration of disease, sex, and rheumatoid factor status did not influence EBV load in patients with RA. In 84 patients with RA, 22 patients with inflammatory arthritides other than RA, and 69 controls, HLA–DR alleles did not seem to influence EBV load. Because we had previously observed that HLA–DRB1*0404 is associated with the lowest and HLA–DRB1*07 with the highest frequencies of T cell precursors to EBV gp110, we compared EBV load in RA patients from these 2 opposite groups. As expected, we found that 18 patients with RA who expressed HLA–DRB1*0404 had higher EBV load (mean 21.95 copies per 150,000 PBMCs) compared with 21 RA patients who expressed HLA–DRB1*07 (mean 7.32 copies/150,000 PBMCs), but this trend did not reach significance (P = 0.08, by t-test). We also observed that shared epitope status did not influence EBV load in patients with RA. In 63 shared epitope–positive patients with RA, the mean EBV copy number was 16.8 copies per 150,000 PBMCs. In contrast, in 20 shared epitope–negative patients with RA, the mean EBV copy number was 11.7 copies per 150,000 PBMCs (P = 0.45, by t-test).
To evaluate whether the level of EBV in the peripheral blood of patients with RA is elevated, we developed an accurate EBV DNA quantitation assay based on real-time PCR. We found that we could detect and quantify EBV DNA in ∼90% of patients with RA, patients with inflammatory conditions other than RA, and controls. However, the number of copies of EBV per 150,000 PBMCs was different between these 3 groups: 16 copies in patients with RA, 6 copies in patients with other inflammatory conditions, and 1.9 copies in normal controls. Thus, patients with RA have an almost 10-fold increase of EBV load in PBMCs when compared with normal controls.
These results are consistent with previous semiquantitative findings by Newkirk et al (13) and Blashke et al (14), who were able to detect EBV DNA in a higher proportion of RA patients than in controls. Recently, real-time PCR was used by Baldanti et al to quantify EBV DNA in the peripheral blood of normal controls and solid-organ transplant recipients. They found that most normal immunocompetent subjects had undetectable levels of EBV DNA in their peripheral blood, but the analysis used DNA from 105 cells, and the amplified fragment was from a unique portion of the EBV genome, thus yielding a sensitivity of detection of 10 copies of EBV (15). The median EBV load in asymptomatic solid-organ transplant recipients was 10 copies/500 ng of DNA, whereas in 15 transplant recipients with symptomatic EBV infection, it was 10,000 copies/500 ng of DNA. There was a “threshold” level of 1,000 EBV copies/500 ng DNA above which patients were likely to develop posttransplant lymphoproliferative disorder. The EBV load we observed in patients with RA is similar to that which is observed in asymptomatic organ transplant recipients and much lower than what is found in symptomatic EBV-infected transplant recipients. Only one patient had EBV load >100 copies/500 ng DNA, but it remained much below the 1,000 copies threshold level.
In another study, Kimura et al (16) used real-time quantitative PCR to quantify EBV in the peripheral blood of patients with infectious mononucleosis, patients with opportunistic B cell lymphomas, and normal controls. The average EBV copy number in 1 μg of PBMC DNA ranged from 15 copies in healthy posttransplant patients to 100 copies in patients with infectious mononucleosis and 5,000 copies in patients with opportunistic B cell lymphomas. Comparing our data with data from those 2 studies suggests that patients with RA have peripheral blood EBV loads similar to those of healthy transplant recipients and much lower than those of immunocompromised, symptomatic EBV-infected transplant recipients.
We studied EBV load in PBMCs of patients with RA and normal controls and observed that it is stable over long periods of time and does not seem to be influenced by methotrexate therapy (7.5–12.5 mg/week) or treatment combining methotrexate (7.5 mg/week) and infliximab (3 mg/kg every 8 weeks). HLA–DR genes do not seem to be a major factor determining EBV load in either patients with RA or controls, and this is somewhat unexpected. Indeed, we had observed that HLA–DRB1*0404 is associated with high frequencies of T cell precursors to EBV gp110 in peripheral blood, whereas HLA–DRB1*07 is associated with low frequencies. EBV gp110 is the EBV equivalent of the gB proteins of herpes viruses; gB proteins are usually critical targets in the control of infection by members of the Herpesvirus family. Nonetheless, the difference in EBV load observed between subjects expressing HLA–DRB1*07 (high responder to gp110, protective against RA) and subjects expressing HLA–DRB1*0404 (low responder to gp110, susceptible to RA) was not significant.
The elevated levels of EBV DNA that we have observed in patients with RA are not a nonspecific consequence of inflammation, because they were not observed in 22 patients with inflammatory arthritides other than RA. However, we do not know whether they are specific for EBV. Indeed, EBV stands out as being the only virus shown to infect almost entire populations of adult humans. We cannot perform any meaningful comparison of viral load in the absence of another virus that would be as widespread as EBV. It is quite possible that the systemic EBV DNA overload we observed in patients with RA might indicate a general deficit in the control of viral infection. Regardless of whether this phenomenon is specific, we believe it is attributable to genes that are associated with RA and are different from HLA–DR.
Our data show that the altered control of EBV infection that was described 20 years ago results in quantifiable (10-fold) systemic EBV overload in patients with RA. Thus, in patients with RA, the Epstein-Barr virus, which is highly recognized by antibodies but never eliminated, is an ideal candidate to cause chronic immune complex disease, and anti-EBV antibody responses should be considered as one of the chronic autoantibody responses that are most relevant to the development of RA (17).