The effect of anti-CD40 ligand antibody on B cells in human systemic lupus erythematosus

Authors


Abstract

Objective

To determine the immunologic effects of anti–CD40 ligand (anti-CD40L) therapy in 5 patients with systemic lupus erythematosus nephritis who participated in an open-label study of a humanized anti-CD40L monoclonal antibody.

Methods

Serum and peripheral blood mononuclear cells were obtained before, during, and after treatment, and the frequency of Ig and anti-DNA antibody–secreting B cells was analyzed by enzyme-linked immunospot assay and by analysis of Epstein-Barr virus (EBV)–transformed B cell lines. To determine the effect of treatment on somatic mutation of Ig genes, reverse transcriptase–polymerase chain reaction was performed on messenger RNA from 4 patients, using primers specific for the DP-47 heavy chain gene and for IgG. Finally, B cell phenotype was investigated using flow cytometry.

Results

Even a brief period of treatment with anti-CD40L markedly reduced the frequency of IgG and IgG anti-DNA antibody–producing B cells, and these changes persisted for several months after cessation of treatment. To confirm these findings, EBV-transformed B cell lines were screened from each of 3 patients, and a 10-fold decrease in anti-DNA antibody–secreting cell lines was found after treatment in all 3 patients. Few differences in mutation patterns were observed before and after treatment; however, the frequency of germline-encoded DP-47 sequences was significantly increased before treatment and normalized following treatment. Flow cytometric analysis of B cells revealed expansion of a CD27−/IgD− B cell subset in some of the patients, which did not change with treatment.

Conclusion

These are the first mechanistic studies of the effect of anti-CD40L therapy in human autoimmune disease. The results suggest that further studies of CD40L blockade are warranted.

The production of the high-affinity anti-DNA antibodies that deposit in the glomerulus of systemic lupus erythematosus (SLE) patients with nephritis requires stimulation of B cells by activated T cells (1, 2). In addition to specific recognition of antigen by T and B cell receptors, lymphocyte activation requires costimulatory signals such as those provided by the CD28/B7 and CD40/CD40L receptor–ligand pairs (3, 4). The interaction of CD40L on activated T cells with CD40 on B cells induces B cell proliferation and formation of germinal centers. Within germinal centers, further cell–cell interactions involving CD40/CD40L lead to B cell maturation through Ig isotype switching, somatic mutation, clonal expansion of high-affinity B cells, and terminal differentiation to plasma cells (4–6). CD40 ligation on memory B cells is also required for their activation and terminal differentiation during secondary antibody responses (7).

Both murine and human SLE are characterized by aberrantly increased expression of CD40L on T cells (8, 9). Blockade of CD40L has been shown to delay the onset of disease in SLE-prone mice and to stabilize or reverse existing renal disease (10, 11). Although the mechanism for this effect is not completely clear, it has been suggested that anti-CD40L may induce long-lasting unresponsiveness of autoreactive B cells to T cell help (10). We have shown in (NZB × NZW)F1 SLE-prone mice that an anti-CD40L monoclonal antibody (mAb) administered to prenephritic mice inhibits both T cell activation and T cell–dependent B cell activation. Following cessation of treatment, activated autoreactive B cells do not emerge for a considerable time despite the ability of the mice to respond to immunization with foreign antigen (Wang X, Davidson A: unpublished observations). We have now had the opportunity to study the mechanism of action of anti-CD40L therapy with respect to B cell function in a few human subjects.

The murine mAb 5c8 effectively blocks human CD40/CD40L interactions in vitro (12); a humanized version of this antibody, hu5c8 (BG9588, Antova; Biogen, Cambridge, MA), has been developed for clinical use. Based on the results of preclinical (13) and phase I studies in patients with refractory immune-mediated thrombocytopenia, suggesting that doses of up to 20 mg/kg were well tolerated, an open-label, multiple-dose study of hu5c8 was initiated in subjects with proliferative SLE glomerulonephritis. Due to the unexpected development of thromboses in a few of the patients, the study was terminated prematurely. We obtained blood samples from 5 study subjects before and after hu5c8 therapy to evaluate the effect of CD40 blockade on anti-DNA antibody–producing B cells. Despite premature termination of the clinical trial because of safety concerns, we found that even a brief period of treatment with hu5c8 markedly reduced the frequency of IgG-producing and IgG anti-DNA antibody–producing B cells, and, in some cases, the B cell immunomodulatory effects persisted for several months after cessation of treatment.

PATIENTS AND METHODS

Study design. Male or female patients ages 18–65 with biopsy-proven World Health Organization class III or IV glomerulonephritis (14) were eligible to enroll. Patients had to satisfy the following criteria: proteinuria >1 gm/24 hours, evidence of hypocomplementemia, elevated titers of anti-DNA antibodies, and hematuria or white cell casts in the urine. Patients were ineligible if they had been treated with high-dose prednisone, cyclophosphamide, mycophenolate mofetil, or cyclosporin A. A total of 30 patients were to be accrued in this multicenter study. The protocol was approved at respective institutional review boards, and study subjects participated after giving informed consent. The hu5c8 (20 mg/kg) was scheduled for intravenous administration on days 1, 15, 29, 57, 85, 113, and 141, and subjects were to be followed up for a subsequent 3 months.

Peripheral blood (10–15 ml) was obtained 4 different times from each of 5 patients enrolled under the supervision of 1 of the authors (RF). The first postdose sample was obtained 15–28 days (day 15–28P) after the patients' final dose, which was 85, 85, 85, 57, and 29 days after the first dose in patients 1–5, respectively. The hu5c8 was still present in the serum at this time because the half-life is 21 days (Biogen: unpublished observations). Subsequent samples were obtained 84 and 168 days after the last dose (days 84P and 168P). A blood sample was obtained from patient 2 250 days after the last dose of hu5c8 because the cells from the pretreatment sample were accidentally lost. Data from the 250P sample were used for the enzyme-linked immunospot (ELISpot) and some of the flow cytometry comparisons for patient 2. Blood samples were transported at 4°C, and lymphocytes were isolated within 3–6 hours in nearly all cases; in no instance was blood used later than 18 hours after being drawn.

Serum studies. Anti–double-stranded DNA (anti-dsDNA) antibodies were measured by Farr assay and by a commercial enzyme-linked immunosorbent assay (ELISA), both of which measure antibodies of all subclasses. In addition, an assay for IgG anti-dsDNA antibodies was performed using a DNA Western-blot technique, as previously described (15). Ig concentrations as well as C3 and C4 levels were measured by ELISA.

Flow cytometry analysis. CD4, CD8, and total B cell numbers before, during, and after treatment were determined for all 5 patients using flow cytometry. In addition, to identify B cell subsets, triple staining was performed on pretreatment, day 84P, and day 168P samples from patients 3, 4, and 5, and on day 84P and day 250P samples from patient 2, with fluorescein isothiocyanate–conjugated anti-CD19, phycoerythrin-conjugated anti-CD27, and anti–IgD biotin (PharMingen, San Diego, CA).

ELISpot assays. Salmon sperm DNA was passed through a Millipore filter (HAWP 45; Millipore, Bedford, MA) to deplete single-stranded DNA and was then coated onto ELISA plates (Falcon, Springfield, NJ) at 10 μg/ml. Plates were allowed to dry, were washed with phosphate buffered saline (PBS), and were blocked with 5% fetal calf serum (FCS)/3% bovine serum albumin (BSA). Two-fold serial dilutions of Ficoll-Hypaque–purified peripheral blood mononuclear cells (PBMCs) were plated in duplicates in Dulbecco's modified Eagle's medium (DMEM)/10% FCS on DNA-coated plates starting at 1–2.5 × 105 viable cells per well. Plates were spun and incubated at 37°C for 2 hours. Alkaline phosphatase–conjugated anti-IgM or anti-IgG (Southern Biotechnology, Birmingham, AL) 1/500 in DMEM/10% FCS was then carefully added, and the plates were incubated overnight at 37°C. Plates were then washed and developed with BCIP (Sigma, St. Louis, MO) 1 mg/ml in AMP buffer (0.75 mM MgCl2 6H2O/0.01% Triton X-405/9.58% 2-amino-2-methyl-1-propanol, pH 10.25).

Spots were counted using a dissecting microscope, by the same observer (AD), who was unaware of the clinical status of the patients, and the frequency of spots per 105 cells was calculated. Total numbers of Ig-secreting cells were measured the same way, using anti-human Ig (1 μg/ml in PBS; BioWhittaker, Baltimore, MD) to coat the plates. A cell line producing an IgG anti-DNA antibody was used as a positive control on each plate. As controls, ELISpot assays were performed the same way on fresh cells from 5 healthy individuals.

Epstein-Barr virus (EBV) transformation. PBMCs were obtained using Ficoll-Hypaque, and 2–3 × 106 cells were transformed by EBV, as previously described (16). Transformed cells were subcloned between days 4 and 7 after transformation and plated at limiting dilutions in 96-well plates. The subcloned lines were then screened for IgM and IgG production and for anti-DNA antibody production using ELISA. Briefly, 96-well ELISA plates were coated with specific goat anti-human IgM or IgG antibodies (1 μg/well in PBS overnight at 4°C) or with 100 μg/ml of dsDNA (prepared as above). After blocking with 5% FCS/3% BSA for 90 minutes, plates were incubated with cell line supernatants for 1 hour at 37°C followed by peroxidase-conjugated goat anti-human IgM and/or IgG antibody (ICN, Costa Mesa, CA) for 1 hour at 37°C and ABTS substrate.

Analysis of somatic mutations of the VH-26 (3-23, DP-47) gene. Total RNA was obtained from 1–2 × 106 PBMCs from pretreatment and the first posttreatment sample and reverse transcribed into complementary DNA (cDNA) using random hexamers (Gibco BRL, Gaithersburg, MD). IgG VH-26 (3-23, DP-47) cDNA libraries were obtained by polymerase chain reaction (PCR) using 5′ VH-26 FR1 primer and a 3′ IgG specific primer, as previously described (17, 18). One round of 35 cycles of PCR was performed using Pfu I polymerase (Stratagene, La Jolla, CA) followed by a 10-minute extension with Taq polymerase (Roche Molecular Systems, Branchburg, NJ). PCR products were subcloned into TOPO TA vector according to the instructions of the manufacturer (Invitrogen, Carlsbad, CA). The colonies from each library were screened for VH-26 using a framework 2 primer as previously described (17).

Positive colonies were then sequenced and compared with the germline VH-26 sequence using a BLAST search. Identical sequences were analyzed only once. Sequences were arbitrarily discarded if there were >25 mutations or >1 run of 3 differences from the germline sequence, excluding triplet deletions. This was done to allow us to compare the sequences with previously described normal panels and accounted for <10% of the obtained sequences. Mutations were analyzed in the V region only, and the last 2 nucleotides of the V region were excluded from the mutation analysis because there were many differences in these nucleotides due to junctional diversity. Ratios of replacement-to-silent mutations were calculated, and the frequency of mutations at the RGYW hot spot on both strands was determined (19). For this analysis, only silent mutations were analyzed to minimize the effect of antigenic selection. Two normal sets of sequences were used for comparison (17, 18). Comparisons of percentage of hot spot mutations in normal versus pretreatment and posttreatment samples were performed using the chi-square test. Finally, the number of transitions and transversions was determined and compared with that found in the normal sequences.

RESULTS

Patient characteristics. All 5 patients met American College of Rheumatology criteria for SLE (20). Demographic data and concurrent immunosuppressive treatments for the patients are shown in Table 1. Patients 1–3 received 4 doses of hu5c8, patient 4 received 3 doses, and patient 5 received 2 doses prior to premature withdrawal from the study. The prednisone dosage remained stable in patients 1, 4, and 5 and was decreased according to protocol in patients 2 and 3. The dosages of other immunosuppressive medications were kept stable throughout the study.

Table 1. Demographic data and concurrent drug therapy of the patients*
PatientAge/sexDuration of SLE, yearsNo. of ACR SLE criteria metDaily prednisone dosageOther treatment
Pre–hu5c8 treatment168P
  • *

    SLE = systemic lupus erythematosus; ACR = American College of Rheumatology; 168P = 168 days posttreatment.

131/F19922
241/M24612.510Hydroxychloroquine, 400 mg/day
335/M35205Azathioprine, 50 mg/day
440/F14555Mycophenolatemofetil, 1,000 mg/day
521/F4544Hydroxychloroquine, 400 mg/day + methotrexate, 15 mg/week

Serologic findings. As shown in Table 2, anti-DNA antibody levels determined by the Farr and IgG radioimmunoassays were elevated prior to treatment with hu5c8 in 4 of the 5 patients studied. Reductions of anti-DNA antibody levels were observed in the initial posttreatment sample (15–28P) in patients 1, 2, and 5; this persisted on day 84 posttreatment in patients 2 and 5. Using the less sensitive commercial ELISA, only patients 1 and 5 had abnormal levels of anti-DNA antibodies at baseline; in both cases, levels decreased after treatment. C3 or C4 levels were below normal in all 5 patients and normalized during or following treatment in patients 1 and 3. Proteinuria decreased in patient 4, who had received only 3 doses of treatment (not shown). Serum IgG levels were unchanged throughout the study period in all 5 patients, and there was a modest increase in serum IgM in 3 of 5 patients (not shown).

Table 2. Serologic evaluation of SLE activity before and after anti-CD40L treatment*
Patient/timeAnti-DNA, unitsComplement, mg/dl
Farr assay, IU/ml (normal <3)ELISA, IU/ml (normal <70)IgG RIA (normal <250)C3 (normal >90)C4 (normal >15)
  • *

    SLE = systemic lupus erythematosus; anti-CD40L = anti–CD40 ligand; ELISA = enzyme-linked immunosorbent assay; RIA = radioimmunoassay; P = posttreatment; ND = not determined.

1
 Pre43.276.91,1667330
 28P28.856.63827032
 84P68.379.65509632
 168P67.8NDND8323
2
 Pre70.744.747854<10
 28P14.221.63386712
 84P33.633.76887213
 168P42.4NDND50<10
3
 Pre3.7251106915
 28P4.021.75013531
 84P3.2186.91,2068317
 168P4.6NDND8415
4
 Pre93.263.36465415
 28P89.553.17045313
 84P104.882.47225413
 168P156.4NDND46<10
5
 Pre1,476.6363.95,40046<10
 15P600.02001,33028<10
 84P466.5169.741637<10
 168P512.2NDND37<10

Flow cytometric findings. Pretreatment total lymphocyte counts and T and B cell numbers are shown in Table 3. Significant abnormalities in B cell numbers and CD4+ T cells were present in 4 of 5 patients. There were no significant changes observed in total T or B cell counts in any patient either during or after treatment (not shown). In order to determine if there were abnormalities in B cell subsets, triple staining was performed with anti-CD19, anti-CD27, and anti-IgD. Four healthy controls and 3 control SLE patients (1 with inactive and 2 with active disease; 1 with nephritis and 1 with mesenteric vasculitis) were also examined. Healthy controls displayed a typical pattern of reactivity with these markers (Figure 1), with the majority of the cells falling into the CD27+/IgD− (class-switched memory), CD27+/IgD+ (IgM memory), and CD27−/IgD+ (naive and B1) subsets (21, 22). The results of this analysis in the SLE patients were highly variable from patient to patient. Only 1 patient with active SLE displayed the phenotype recently described by Odendahl et al, in which naive B cells were virtually absent (23). In 2 of the SLE patients, a large population (20–40% of CD19+ cells) of CD27−/IgD− B cells was observed. This was particularly striking in patient 5 but was also observed in patient 3 and in 2 of the control SLE patients (Figure 1). Except for patient 5, in whom the percentage of CD27high cells decreased 3-fold by the end of treatment, the B cell subset profile defined by the markers we used did not change substantially during or after therapy (data not shown).

Table 3. Lymphocyte subsets prior to treatment
PatientCD19+,/mm3 (normal >200)CD4+,/mm3 (normal >500)CD8+,/mm3 (normal >300)Total,/mm3 (normal >1,200)
1908194751,540
238199136420
363269162630
44873475251,710
5176285155930
Figure 1.

Ficoll-Hypaque–purified peripheral blood mononuclear cells (PBMCs) stained with anti-CD19, anti-CD27, and anti-IgD. Cells were gated on CD19. A, Control patient with inactive systemic lupus erythematosus (SLE). B, Control patient with active SLE and mesenteric vasculitis. C, Healthy control. D, Patient 2, day 250P. E, Patient 5, day 0. A large CD27−/IgD− population was observed in the patient with inactive SLE and in patients 4 (not shown) and 5.

ELISpot results. The ELISpot assay detects individual B cells that spontaneously produce IgM or IgG. Despite the unchanged total B cell numbers following treatment, marked differences were observed in the frequency of Ig- or anti-DNA antibody–producing B cells revealed by ELISpot (Table 4). The frequencies of both IgM- and IgG-producing B cells were reduced in patients 2–5 at the first or second posttreatment visit; the decrease in IgM-producing cells was 2–11-fold, whereas the decrease in IgG-secreting cells was 5–25-fold. The number of Ig-secreting cells returned to baseline by day 168P. Anti-DNA antibody–producing B cells also decreased markedly during treatment and were below the level of detection in 4 of 5 patients at the first or second posttreatment visit, representing decreases of 20–140-fold.

Table 4. Frequency of total Ig and Ig anti-DNA–secreting cells by enzyme-linked immunospot assay*
Parameter, subjectPre–hu5c8 treatment15–28P84P168P
  • *

    Values are per 105 cells. For total IgG, P < 0.02, SLE patients versus controls, pretreatment mean; for IgG anti-DNA, P < 0.01, SLE patients versus controls, pretreatment mean (by Mann-Whitney U test). There were too few paired samples for valid statistical comparison of pre- and posttreatment samples. U = no spots detected in any well (limit of detection 0.25/105 cells). See Table 2 for other definitions.

  • Data from day 250P sample.

  • Spots detected at >0.25 but <1/105 cells.

Total IgM
 Patient 115141640
 Patient 2125744
 Patient 31561016
 Patient 48026660
 Patient 5405270180
 Controls, mean ± SD 20 ± 9.5
Total IgG
 Patient 12641004480
 Patient 21282014160
 Patient 323415100900
 Patient 424013050350
 Patient 52,400942601,700
 Controls, mean ± SD 113 ± 52
IgM anti-DNA
 Patient 11.61<1<1
 Patient 2171
 Patient 33U43.5
 Patient 42.5U11
 Patient 53<1<17
 Controls, mean ± SD 1.6 ± 1.7
IgG anti-DNA
 Patient 111211
 Patient 25UU1
 Patient 322U124
 Patient 45U41
 Patient 536<1U2
 Controls, mean ± SD 0.3 ± 0.38

EBV-transformed cell lines. The decrease in frequency of anti-DNA antibody–producing B cells was paralleled by the findings on analysis of EBV-transformed B cell lines before treatment and at the first posttreatment visit. EBV binds to the C3d receptor that is present on mature B cells and on early plasma cells (24, 25); thus, the population of EBV-transformed cells may differ from that detected by ELISpot. Successful transformations at both pretreatment and posttreatment times were obtained in 3 patients, and >95% of the lines secreted Ig, predominantly of the IgM and IgG isotypes, with no differences in the relative frequency of these isotypes before and after treatment. The lines were screened by ELISA for dsDNA binding. In each case, there was a decrease in the frequency of cell lines that displayed binding to dsDNA in the posttreatment sample (Table 5).

Table 5. Frequency of DNA binding lines among EBV transformants*
Patient, timeNumber screenedPositive anti-DNA, number (%)
  • *

    P < 0.0001, frequency of DNA binding lines, pretreatment mean versus posttreatment mean (by Mann-Whitney U test). EBV = Epstein-Barr virus; Pre = pre–hu5c8 treatment; P = posttreatment.

  • Isotyping of positive lines was performed on 13 pretreatment lines (6 IgG, 7 IgM) and 4 posttreatment lines (1 IgG, 3 IgM).

3
 Pre1529 (5.9)
 28P1763 (1.7)
4
 Pre28119 (6.7)
 28P3682 (0.5)
5
 Pre363 (8.3)
 15P1760 (0)

Somatic mutations. Since anti-CD40L treatment has been shown to disrupt germinal centers and decrease the frequency of somatic mutations (26), we sought to determine whether we could detect alterations in the frequency of somatic mutations of the VH-26 gene that is expressed among autoreactive antibodies including anti-DNA antibodies. VH-26 has been extensively analyzed in healthy individuals, and 2 normal sets of sequences are available for comparison (17, 18). To obtain a sufficient number of mutated genes to analyze, we sampled the activated or memory B cell pool, represented by IgG antibodies. This was done by performing reverse transcriptase–PCR (RT-PCR) on RNA using a 3′ IgG primer to select the class-switched genes. PCR products from pretreatment and the first posttreatment samples were successfully obtained from patients 2–5. As expected, less PCR product was obtained in each case from the posttreatment sample. Forty-three pretreatment and 34 posttreatment sequences were analyzed (Figure 2 and Table 6). Surprisingly, large numbers of sequences containing ≤3 mutations were obtained from the pretreatment samples of patients 2, 3, and 4, constituting almost 50% of the sequences obtained from these patients (Figure 2). This did not occur in healthy individuals, from whom only 10% of sequences had ≤3 mutations (P < 0.001).

Figure 2.

Mutations per sequence in VH-26 sequences obtained from healthy control individuals (67 sequences) and pretreatment systemic lupus erythematosus (SLE) individuals (43 sequences), shown as the percentage of sequences containing the given frequency of mutations. Only 10% of sequences from healthy individuals had ≤3 mutations, compared with 48% of those from SLE individuals (P < 0.001, healthy subjects versus SLE patients, by chi-square analysis). The inset table shows the frequency of sequences containing ≤3 mutations for each of the patients before and after treatment (P < 0.02 versus pretreatment for patients 2 and 3, by chi-square analysis).

Table 6. Mutation analysis*
GroupNo. sequences analyzedTotal mutationsCDR R:SFramework R:SSilent mutations
Hot spot*Non–hot spot
  • *

    No significant differences were seen between any of the groups, by chi-square analysis. CDR = complementarity-determining regions; R:S = ratio of replacement-to-silent mutations.

  • *

    Hot spots counted for this analysis are RGYW (R = A/G; Y = T/C; W = A/T) on either strand.

Healthy1215565:1145:341530
Pretreatment43261100:2574:622067
Postteatment34274110:2684:542159

After treatment, these types of sequences disappeared in 2 of 3 patients but continued to dominate the repertoire in the remaining patient (Figure 2). In order to confirm that this abnormality also occurred for other VH genes, we also performed RT-PCR on pretreatment samples for the nonpolymorphic VH6 gene. This gene is less abundantly expressed, and we were only able to obtain PCR products from patients 2 and 5. Twelve sequences from patient 2 were analyzed and we again found that 4 of 12 (33%) had ≤2 mutations (not shown). No other differences in the mutation pattern were found between treated patients and controls or between pretreatment and posttreatment samples (not shown).

DISCUSSION

Glomerulonephritis occurs in at least 40% of patients with SLE and is currently treated with corticosteroids and other immunosuppressive agents that are associated with significant morbidity and do not prevent subsequent disease flares in up to 50% of treated patients (27). Safer and more effective therapies for SLE nephritis are clearly needed.

Studies in murine models of SLE have shown that costimulatory blockade, including the use of antagonists of the CD40/CD40L interaction, may be a promising new approach to the treatment of autoimmune disease. In these animal models, anti-CD40L mAb is effective in delaying disease onset and has some beneficial effects, even in the face of established nephritis (10, 11). The current clinical study was designed to determine whether a similar strategy could be applied to human disease. The anti-CD40L mAb studied, hu5c8, was used at a dosage that appeared efficacious and safe in preclinical studies, but was unexpectedly found to be associated with thrombotic events in 2 patients, necessitating premature termination of the clinical study (28). Although the duration of hu5c8 therapy was limited and blood samples were available from only a few patients, we were able to draw several conclusions regarding the mechanism of action of anti-CD40L antibody in humans.

SLE patients demonstrate profound disturbances in B cell homeostasis (23, 25). Patients with both active and inactive SLE can have markedly diminished peripheral B cell numbers, accounted for predominantly by a decrease in naive CD27− B cells, which represent 40–60% of peripheral B cells in healthy individuals (21, 23). Despite this, the relative frequency of activated B cells is generally increased in SLE patients (29). Accordingly, 4 of 5 patients had reduced numbers of peripheral B cells (Table 3), but in all 5, the frequency of B cells spontaneously secreting IgG was increased (P < 0.02) (Table 4).

This is consistent with previous findings that SLE patients have a marked increase in the frequency of activated CD27high or CD38high plasma cells in the peripheral blood compared with healthy controls (23, 25). The 4 patients studied by flow cytometry varied in the proportions of class-switched memory cells (CD27+/IgD−), IgM-producing memory B cells (CD27+/IgD+), and naive and CD5+ B cells (CD27−/IgD+). Two of the treated SLE patients as well as 2 of 3 SLE controls had a large population of CD27−/IgD− cells that were absent from healthy controls. These might represent either class-switched memory cells that have down-regulated CD27 (30) or, alternatively, class-switched cells from the B1 or marginal zone subsets.

The most striking effect of hu5c8 administration was the decrease in 4 of 5 patients in the frequency of B cells spontaneously secreting Ig. This effect was associated with a marked decrease in the frequency of anti-DNA antibody–secreting cells of both IgM and IgG isotypes in the peripheral blood, with a slow return to baseline after discontinuation of treatment. The effect on B cells of the IgG isotype was more profound and lasted longer than the effect on IgM-producing cells. These findings suggest that the release of activated antibody-secreting cells into the peripheral blood, either from newly recruited cells or from the memory compartment, was influenced by anti-CD40L therapy. This is consistent with observations both in rodents and primates that germinal center formation and memory cell activation are attenuated in the presence of anti-CD40L treatment (31), and with observations by Grammer et al in another small cohort of hu5c8-treated SLE patients, showing a decrease in CD38high-activated plasma cells in the peripheral blood after treatment (32). In addition, in 4 of the patients there appeared to be a more profound effect on IgG anti-DNA antibody–producing B cells than could be accounted for by the decrease in total numbers of IgG-secreting cells, suggesting that autoreactive B cells may be preferentially sensitive to anti-CD40L therapy. We have noted similar findings in the spleens of mice treated with the costimulatory blocker CTLA-4Ig (33).

Despite the decrease in frequency of total IgG-secreting cells by 50–90% and of anti-DNA antibody–secreting B cells in the peripheral blood to below the level of detection in 4 of 5 patients, there was no change in serum IgG levels, and the levels of anti-DNA antibodies in the serum did not decrease to normal. This suggests that while the antibody-secreting cells in the peripheral blood that have the phenotype of early plasma cells (25) or their germinal center precursors are sensitive to CD40L blockade, at least some of the anti-DNA antibodies in the serum derive from B cells that are not sensitive to CD40L blockade. These cells could include long-lived bone marrow plasma cells that require neither T cells nor the continuing presence of antigen to survive and secrete antibody (34).

Alternatively, there may be a subpopulation of B cells that is able to mature and secrete IgG in autoimmune individuals even in the absence of costimulation. We have observed a similar maintenance of serum IgG levels and of frequencies of bone marrow–derived antibody-producing cells despite profound decreases in spleen-derived antibody-producing cells in rodents treated with a different costimulatory blocker, CTLA-4Ig (33). Following cessation of therapy, spontaneous anti-DNA antibody–secreting B cells were again found in the peripheral blood, showing that CD40L blockade for periods up to 85 days had not induced tolerance.

Analysis of the well-described VH-26 (DP-47) gene was undertaken to determine whether any abnormalities in the frequency or pattern of somatic mutations of this gene could be observed in the study patients. Previous studies of peripheral B cells using very small numbers of SLE patients have suggested that SLE patients may have hypermutated V region genes and abnormalities in hot spot targeting (35, 36). Patients deficient in CD40 or its ligand have a defect in hot spot targeting (37). Using the VH-26 gene as a marker, we were unable to detect any overall major differences in mutation frequency and pattern in peripheral B cells of the SLE patients before or after treatment compared with healthy controls (Table 6), although there was some variability among patients.

Our mutation analysis did, however, identify what appears to be a novel B cell subset in 3 of the 4 SLE patients studied. Analysis of the VH-26 gene revealed a population of B cells that had undergone class switching to IgG but displayed few or no somatic mutations. This abnormal population of B cells disappeared in 2 of the 3 patients following anti-CD40L treatment. The nature of this B cell subset remains to be elucidated. One possibility is that these unmutated V regions might have recently undergone receptor editing. However, in both germline- and non–germline-encoded V regions there was a similar distribution of J regions, with marked overrepresentation of JH4 in both sets (data not shown). An alternative hypothesis is that these cells derive from non–germinal center marginal zone B cells or B1 cells, increases in which have been described in murine models of SLE (38, 39). In SLE-prone mice, these subsets contain a markedly increased frequency of DNA binding cells. It is of interest that Gur et al have also described a decrease in the frequency of CD5bright B cells in the peripheral blood following treatment in their group of patients treated with hu5c8 (40).

The potential role of CD40L blockade in the treatment of SLE remains to be defined. The efficacy of hu5c8 has been analyzed in 18 patients who received 3 or more doses of treatment; improvements were noted in serologic markers and in hematuria. One month following treatment, the overall decrease in anti-DNA antibodies in the group was 38.9% (28), consistent with what was observed in our 5 patients. Anti-CD40L agents are extremely powerful inhibitors of T cell–dependent B cell activation in prevention studies in murine SLE (41, 42), and when used in high doses are able to stabilize or reverse disease even in the setting of active nephritis (11). Whether the immediate benefits observed in mice with established disease are due to a direct effect on pathogenic autoantibodies or whether they are due to blockade of other functions influenced by CD40/CD40L interactions, including activation and migration of cells involved in the inflammatory response (31, 43), remains to be determined.

Since it directly measures the effect of therapy on activated autoreactive B cells released into the peripheral blood, the ELISpot assay may be useful as a biomarker of response of this component of immune activation to therapy. Analysis of multiple SLE patients will be required to determine the range of abnormal B cell phenotypes associated with the disease and their responsiveness to new treatment modalities. Development of alternate approaches to block CD40L with a less toxic agent and trials of combination therapies, either with conventional treatment or other biologic agents (44), appear warranted.

Acknowledgements

The authors acknowledge the expert assistance of study coordinator Laurie Payne, North Shore University Hospital. We would also like to thank Drs. B. Diamond and H. Keiser for critical review of the manuscript.

Ancillary