Rituximab improves peripheral B cell ... - Wiley Online Library

ARTHRITIS & RHEUMATISM Vol. 50, No. 11, November 2004, pp 3580–3590 DOI 10.1002/art.20592 © 2004, American College of Rheumatology

Rituximab Improves Peripheral B Cell Abnormalities in Human Systemic Lupus Erythematosus Jennifer H. Anolik,1 Jennifer Barnard,1 Amedeo Cappione,1 Aimee E. Pugh-Bernard,2 Raymond E. Felgar,1 R. John Looney,1 and In ˜aki Sanz1 tion. The frequency of autoreactive VH4.34 memory B cells also decreased 1 year posttreatment, despite the presence of low levels of residual memory B cells at the point of maximal B cell depletion and persistently elevated serum autoantibody titers in most patients. Conclusion. This study is the first to show evidence that in SLE, specific B cell depletion therapy with rituximab dramatically improves abnormalities in B cell homeostasis and tolerance that are characteristic of this disease. The persistence of elevated autoantibody titers may reflect the presence of low levels of residual autoreactive memory B cells and/or long-lived autoreactive plasma cells.

Objective. B lymphocyte depletion has recently emerged as a promising approach to the treatment of systemic lupus erythematosus (SLE). As part of a phase I/II dose-ranging trial of rituximab in the treatment of SLE, we evaluated the fate of discrete B cell subsets in the setting of selective depletion by anti-CD20 monoclonal antibody and during the B cell recovery phase. Methods. B cell depletion and phenotype were examined by flow cytometry of peripheral blood mononuclear cells for CD19, CD20, CD27, IgD, and CD38 expression. Changes in autoreactive B lymphocytes and plasma cells were assessed by determination of serum autoantibody levels (anti–double-stranded DNA and VH4.34) and by direct monitoring of a unique autoreactive B cell population bearing surface antibodies whose heavy chain is encoded by the VH4.34 gene segment. Results. Compared with normal controls, SLE patients displayed several abnormalities in peripheral B cell homeostasis at baseline, including naive lymphopenia, expansion of a CD27ⴚ,IgDⴚ (double negative) population, and expansion of circulating plasmablasts. Remarkably, these abnormalities resolved after effective B cell depletion with rituximab and immune reconstitu-

Systemic lupus erythematosus (SLE) is an autoimmune disease with heterogeneous clinical manifestations characterized by the generation of pathogenic autoantibodies directed against chromatin and a variety of other nuclear antigens. Although the pathogenesis of SLE is not yet fully understood, a growing body of experimental evidence indicates that B lymphocytes play a central role. In SLE, B cells may disturb immune homeostasis by multiple mechanisms in addition to the production of pathogenic autoantibodies, including autoantigen presentation, cytokine production, and modulation of the T cell repertoire and T cell memory (1–4). Overall, the implication is that loss of B cell tolerance is likely critical to SLE disease pathogenesis. The precise interplay between genetic defects and environmental influences that must underlie this loss of tolerance and subsequent disease progression remains to be fully elucidated. Investigation of the genetics of lupus in mice and humans suggests the importance of defects in apoptosis, immune complex clearance, and lymphoid signaling (5–7). Defects in lymphoid signaling may include defects that lower the activation threshold of B cells and lead to B cell hyperactivity and immune dysregulation. In human SLE the evidence for B cell

Dr. Anolik’s work was supported by the NIH (grant K08-AR048303 from the National Institute of Arthritis and Musculoskeletal and Skin Diseases) and the Lupus Foundation of America. Dr. Sanz’ work was supported by the NIH (grants RO1-AI-049660-01A1 and U19-AI-56390 from the National Institute of Allergy and Infectious Diseases). 1 Jennifer H. Anolik, MD, PhD, Jennifer Barnard, MS, Amedeo Cappione, PhD, Raymond E. Felgar, MD, PhD, R. John Looney, MD, In ˜aki Sanz, MD: University of Rochester School of Medicine and Dentistry, Rochester, New York; 2Aimee E. PughBernard, PhD: National Jewish Medical and Research Center, Denver, Colorado. Address correspondence and reprint requests to Jennifer H. Anolik, MD, PhD, University of Rochester School of Medicine, Box 695, 601 Elmwood Avenue, Rochester, NY 14642. E-mail: [email protected]. Submitted for publication April 6, 2004; accepted in revised form July 26, 2004. 3580

B CELLS IN SLE AFTER RITUXIMAB

hyperactivity is multifold and includes the presence of increased numbers of spontaneous immunoglobulinsecreting peripheral B cells, increased calcium flux upon signaling through the B cell receptor, and expression of high levels of costimulatory molecules CD80, CD86, and CD40 ligand on B cells (8,9). Additionally, recent evidence suggests a role in SLE for high serum levels of B lymphocyte stimulator (BLyS), a member of the tumor necrosis factor family of cytokines that promotes B cell maturation and survival and plasma cell differentiation (10,11). BLyS, along with other cytokines, intrinsic B cell defects, and the abnormal influence of other immune cells, may contribute to the defects in peripheral blood B cell subpopulations that have been observed in SLE. These include a naive B cell lymphopenia, circulating germinal center (GC) founder B cells, and expansion of circulating plasmablasts (12–14). Given the critical role of B cells in the pathogenesis of SLE, we hypothesized that the targeted elimination of B cells has the potential to induce long-lasting remissions and reestablish B cell tolerance. Rituximab is a chimeric mouse–human monoclonal antibody against the B cell–specific antigen CD20, which depletes B lymphocytes in vivo from the pre-B stage in the bone marrow, when CD20 is first expressed, to the mature B cell stage. Because CD20 is not expressed on early bone marrow B cell precursors and plasma cells, these cells are not susceptible to depletion with rituximab. Rituximab represents an effective treatment of B cell lymphomas and has emerged as a promising potential treatment of several autoimmune diseases in which B cells may play an important role, including lupus (15–19). We recently reported results of a phase I/II dose-escalation trial in which rituximab-induced B cell depletion significantly improved SLE clinical disease activity (18,20). As a first examination of the immunologic effects of rituximab in SLE, we evaluated B cell subsets and autoantibodies in patients at baseline, in the setting of selective depletion by anti-CD20 monoclonal antibodies (mAb), and during the B cell recovery phase. This is the first mechanistic study of the effects of B cell depletion on B cell abnormalities in human SLE. Our results indicate that targeted B cell depletion effectively normalizes many of the significant disturbances in peripheral B lymphocyte homeostasis that are characteristic of active SLE. PATIENTS AND METHODS Patient samples and study design. Eligible study subjects were men and women, age ⱖ18 years, who met the

3581

American College of Rheumatology criteria for the classification of SLE (21) and had clinically active disease as defined by a Systemic Lupus Activity Measure (SLAM) index score ⬎5 (22). Patients receiving cyclophosphamide or high-dose prednisone were excluded, but other immunosuppressive medications were allowed if the regimen had been stable for at least 1 month prior to study entry (see Table 1 for patient characteristics). Detailed informed consent was obtained from all patients and healthy donors, in accordance with protocols approved by the Human Subjects Institutional Review Board of the University of Rochester Medical Center. Rituximab was provided by Genentech (South San Francisco, CA) and IDEC Pharmaceuticals (San Diego, CA). The dose-escalation protocol was used, as follows: low dose ⫽ 1 infusion of 100 mg/m2 (first 6 subjects), intermediate dose ⫽ 1 infusion of 375 mg/m2 (next 6 subjects), and high dose ⫽ 4 weekly doses of 375 mg/m2 (final 6 subjects). Patient 13 withdrew from the study prior to receiving the study drug, and patient 19 moved out of state following the first infusion. The remaining 17 patients were available for analysis. Patient 17 received only 1 infusion of 375 mg/m2 and therefore was included in the intermediate-dose group for analysis. Peripheral blood lymphocytes (PBLs) were isolated from heparinized blood by Ficoll-Hypaque density-gradient centrifugation (Pharmacia Biotech, Uppsala, Sweden). In select experiments (when analyzing residual B cells at the point of maximal depletion), B cells were purified by CD19⫹ magnetic selection or magnetic negative selection using a hapten antibody and anti-hapten microbeads (MACS indirect B cell isolation kit) (Miltenyi Biotec, Bergisch Gladbach, Germany). Flow cytometry. Immunofluorescence staining for flow cytometric analysis was performed by incubating PBLs with excess mAb in phosphate buffered saline (PBS)/1% bovine serum albumin (BSA) on ice for 20 minutes after blocking with 10 ␮g human IgG for 20 minutes, or by standard direct staining and lysing/washing methods, as previously described (20). Cells were washed in PBS/BSA and in some cases incubated with streptavidin–peridin chlorophyll protein. Cells were then washed and fixed in PBS/1% paraformaldehyde before analysis on a FACSCalibur flow cytometer (Becton Dickinson, Mountain View, CA). B cells were identified based on CD19 expression; CD20 expression was also examined. In all patients, CD19⫹ B cell percentages were assessed at baseline and at 1, 2, 3, 6, 9, and 12 months following rituximab infusion (in some instances a later time point [18–36 months] was also evaluated). Absolute B cell numbers were calculated based on the white blood cell count, the percentage of lymphocytes, and the percentage of CD19 cells identified on flow cytometry. Effective B cell depletion was defined as depletion of B cells to ⬍1% of the total peripheral blood lymphocytes (absolute cell count ⬍5 cells/␮l). Additional phenotypic analysis with IgD/CD27 and/or IgD/CD38 and/or CD19/CD38, as described below, was performed in a subset of patients at baseline (n ⫽ 11 [patients 1, 8–12, 14, 15–18]) and in all patients at various postinfusion time points. For consistent comparison at different time points, B cell subsets were quantitated as a percentage of total peripheral blood B cells, using Ficoll-isolated PBLs and identification of the B cells with CD19. Naive (CD27⫺,IgD⫹) and isotype-switched memory (CD27⫹,IgD⫺) B cell subsets were

3582

ANOLIK ET AL

Table 1. Patient characteristics*

Patient 1 2 3 4 5 6 7 8 9 10 11 12 14 15 16 17 18

Age/sex/race 37/F/C 45/M/C 44/F/C 41/F/C 29/F/A 39/F/B 39/F/C 30/F/B 22/F/H 47/F/C 43/F/C 41/F/C 30/F/C 31/F/C 37/F/B 29/F/B 46/F/B

Disease duration, years

Daily prednisone dose, mg

12 10 23 3 5 15 17 8 3 10 17 7 5 14 15 17 10

16 15 15 10 3 15 7 0 15 4 30 10 10 5 25 40 7.5

Other treatment

Nephritis†

SLAM

Rituximab dose

Depletion‡

IV IV III – IV – III – III – – ⫹ – – – – –

13 11 10 8 6 8 6 13 9 6 10 8 6 9 11 12 6

Low Low Low Low Low Low Medium Medium Medium Medium Medium Medium High High High Medium High

⫹ ⫹ – ⫹ ⫹ – – – ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ – –

– AZA – HCQ, AZA HCQ, AZA HCQ AZA MYC HCQ, AZA HCQ AZA HCQ AZA AZA MTX HCQ HCQ

* SLAM ⫽ Systemic Lupus Activity Measure; C ⫽ Caucasian; AZA ⫽ azathioprine; HCQ ⫽ hydroxychloroquine; A ⫽ Asian; B ⫽ black; H⫽ Hispanic; MYC ⫽ mycophenolate; MTX ⫽ methotrexate. † Presence or absence and, when known, type. ‡ ⫹ ⫽ depletion of B cells to ⬍1% of the total peripheral blood lymphocytes or an absolute cell count of ⬍5 cells/␮l.

identified on the basis of their expression of IgD and CD27 (n ⫽ 8 at baseline; n ⫽ 15 at followup) (23,24). Staining for CD38 and IgD classified B cells along the developmental stage from naive to memory, which is the mature B cell (Bm) classification (Bm1–Bm5) originally described for tonsillar B cell subpopulations (25,26). The Bm2⬘ or pre-GC population was also identified as CD38high,CD19⫹,CD20⫹. Plasmablasts were identified as CD38 high ,CD19 low ,CD20⫺ or CD38high,IgD⫺,CD20⫺. The phenotypic summary of these populations is provided in Table 2. In a subset of patients at baseline (n ⫽ 6 [patients 1, 9, 10, 12, 14, and 15]) and in all patients at followup, autoreactive VH4.34 B cells were specifically identified using

a rat monoclonal antiidiotype (9G4) that recognizes a conserved determinant encoded by the framework 1 region of VH4.34 antibodies. Expression of the 9G4 idiotype is essentially synonymous with autoreactivity, and its detection can be used to track autoreactive VH4.34 cells in the GC and memory compartments (23). Serum autoantibodies. Anti–double-stranded DNA (anti-dsDNA) antibodies (IgG and IgM) were measured by a commercial enzyme-linked immunosorbent assay (ELISA) at all time points. VH4.34 antibodies were detected by ELISA at baseline, 12 months, and at longer followup time points (up to 36 months), as follows: Nunc C plates (Nunc, Rochester, NY) were coated with purified 9G4 antibody at 1 ␮g/ml in sodium

Table 2. Summary of surface marker expression on B cells* Pro B

Pre-B

Immature

Transitional

Mature naive (Bm1/2)

Pre-GC (Bm2⬘)

GC (Bm3/4)

Memory (Bm5)

PC†

CD19 CD20⫺‡ CD38high CD10high CD24high

CD19 CD20 CD38high CD10high CD24high

CD19 CD20 CD38high CD10int CD24high IgM

CD19 CD20 CD38high CD10⫺ CD24high IgMhigh IgD⫺/low

CD19 CD20 CD38⫺/low CD10⫺ CD24int IgM IgD

CD19 CD20 CD38high CD10

CD19 CD20 CD38high CD10

CD19 CD20 CD38⫺/low

CD19low CD20⫺ CD38high

IgD⫺ CD27

IgD⫺§ CD27

IgD⫺ CD27

IgM IgD

CD23 (Bm2) CD2133333 increasing expression * GC ⫽ germinal center; Bm ⫽ mature B; PC ⫽ plasma cell; int ⫽ intermediate. See refs. 1, 3, 25, 26, and 51. † The indicated markers are characteristic of peripheral blood plasmablasts. The expression of several markers, including CD27 and CD38, varies with the tissue source and maturity of the plasma cell. ‡ All cells are negative for this marker. § A fraction of peripheral blood memory cells are IgD⫹/IgM⫹ (preswitch). The majority of peripheral blood memory cells are CD27⫹, IgD⫺ postswitch and express either surface IgG, IgA, or IgE.


3583

carbonate buffer for 1.5 hours at 37°C. The plates were blocked with 0.25% PBS/BSA. Sera were serially diluted in 0.25% BSA/Hanks’ balanced salt solution starting with 1:1,000 for VH4.34 IgG and 1:20 for VH4.34 IgM. Binding was detected with goat anti-human IgG or IgM, respectively, conjugated to alkaline phosphatase. VH4.34 IgG or IgM purified from high-titer patient’s serum were included as standards on each plate. Levels greater than 3 SDs above the normal mean were considered elevated for each isotype and were expressed relative to total IgG or IgM as determined by nephelometry (14). Statistical analysis. Statistical significance was assessed by 2-tailed Student’s t-test using Excel software (Microsoft, Redmond, WA) or by Pearson’s correlation using StatQuest software (Stata Corporation, College Station, TX). P values less than or equal to 0.05 and R2 values greater than 0.5 were considered significant.

RESULTS Abnormalities in peripheral B cell homeostasis in SLE. Peripheral blood B cells were initially examined for the expression of CD27 and IgD in order to differentiate naive and memory B cells and plasmablasts according to accepted classification criteria (24) (see Table 2). Using this approach, several studies, including our own (12–14,27), had previously identified 2 consistent abnormalities in patients with active SLE: 1) naive lymphopenia with a relative increase in memory cells, and 2) expansion of CD27high plasmablasts. In the current study, data for 8 SLE patients (at baseline) and 7 normal controls were analyzed in this manner. Our results confirmed the presence of a prominent reduction in naive IgD⫹,CD27⫺ B cells in the peripheral blood of SLE patients before treatment with rituximab compared with normal controls (35 ⫾ 17% versus 68 ⫾ 6%; P ⫽ 0.0008) (Figure 1). Consistent with our previous findings (14), we detected an association between naive lymphopenia and autoantibody titers (for VH4.34, R2 ⫽ 0.6, P ⫽ 0.05). As shown in Figures 1A and 2, patients with SLE also displayed a significant expansion of peripheral blood CD27high plasmablasts, which was better identified by the inclusion of CD38 staining (CD38high, CD19low,CD20⫺). Thus, plasmablasts were significantly expanded in the SLE cohort at baseline (n ⫽ 15) compared with normal controls (mean ⫾ SD 18.5 ⫾ 17.9% [range 0–51%] versus 0.24 ⫾ 0.23% [range 0–0.5%]; P ⫽ 0.001) (Figures 2A and B). CD38 staining also permits detection of a distinct GC founder population (pre-GC: CD38high,CD19⫹,CD20⫹), which has been previously reported to be expanded in SLE (13). In our cohort, expansion of plasmablast pre-GC cells

Figure 1. Normalization of peripheral blood B cell–naive lymphopenia and double-negative expansion after effective B cell depletion with rituximab (Rx) in patients with systemic lupus erythematosus (SLE). A, Peripheral blood B cells (CD19⫹) (1 ⫻ 106) were examined by flow cytometry for the expression of CD27 and IgD. This analysis is depicted for a normal control and for SLE patient 14 before treatment with high-dose rituximab and after immune reconstitution (20-month time point). The percentage of B cells in each of the 4 quadrants, representing naive (N), memory (M), double-negative (DN), and double-positive (preswitch memory) populations are shown. The CD27high plasmablast population (PC) is depicted in the circled gate. B, The percentage of IgD⫹,CD27⫺ naive cells is depicted (based on flow cytometry analysis as in A) for 7 normal controls, 8 SLE patients before rituximab therapy, and 15 SLE patients after immune reconstitution (ⱖ1 year after receiving rituximab); the means for each group are indicated by the horizontal lines. The circled points represent patients with relatively ineffective B cell depletion after rituximab. P values were determined by Student’s t-test.

(mean ⫾ SD 4.7 ⫾ 2.8% in normal controls) was observed in only 3 of 15 patients studied (for the entire cohort, 9 ⫾ 11.9% [range 0–35%]) (as shown in Figure 2A for patient 18). Our baseline SLE studies also corroborated the expansion of a B cell population previously described by us and other investigators (28,29), characterized by the absence of both CD27 and IgD (double-negative cells). Double-negative cells were significantly expanded in the peripheral blood of SLE patients at baseline compared with normal controls (mean ⫾ SD 14.4 ⫾ 7.9% versus 3.9 ⫾ 1.9%; P ⫽ 0.01) (Figure 1A, and data not shown). The identity of the double-negative population is cur-

3584

Figure 2. Improvement of peripheral plasma cell (PC) expansion after B cell depletion with rituximab (Rx) and immune reconstitution in patients with systemic lupus erythematosus (SLE). A, Total peripheral blood lymphocytes (PBLs) were examined for expression of CD38 and CD19. The CD19⫺ populations represent non–B cells. CD19⫹ B cells show a range of CD38 expression, with the highest expression found on pre–germinal center (pre-GC) cells (rectangle) and plasmablast/PCs (circle; CD19low). In the dot plots of PBLs from 2 SLE patients, the red gate (where present) represents CD20⫹ cells. Patient 14 had a significant PC expansion (CD19low,CD20⫺,CD38high), which decreased after treatment with high-dose rituximab and immune reconstitution (20-month time point). Patient 18 also had a significant pre-GC expansion (CD19⫹,CD20⫹,CD38high and CD10⫹ [data not shown]), which increased after high-dose rituximab (shown for the 3-month time point; results were similar at 1 year). The indicated subpopulation percentages are expressed relative to the total number of peripheral blood B cells. B, Percentage of PCs for 5 normal controls, 15 SLE patients before rituximab treatment, and 15 SLE patients ⱖ1 year after rituximab; the means for each group are indicated by the horizontal lines. The percentage of PCs was calculated based on flow cytometry analysis as shown in A and as described in Patients and Methods. The circled point represents patient 6, who had ineffective B cell depletion. P values were determined by Student’s t-test. C, Analysis of patient 11 before and 2 months after intermediate-dose rituximab treatment, showing rapid resolution of PC expansion (0.22 cells/␮l at 2 months [2 mo.] versus 3.59 cells/␮l at baseline).

rently under investigation. As demonstrated for naive lymphopenia, there was a statistically significant correlation between the degree of double-negative expansion

ANOLIK ET AL

and higher autoantibody titers (for VH4.34 autoantibodies, R2 ⫽ 0.8, P ⫽ 0.05). Effect of B cell depletion on abnormalities in peripheral B cell homeostasis in SLE. After treatment with rituximab and immune reconstitution (ⱖ1 year posttreatment), there was a statistically significant increase in the percentage of naive B cells for the group as a whole (P ⫽ 0.03); this increase was even more striking (P ⫽ 0.008) for the 11 patients with effective B cell depletion (see below) (Figure 1A, patient 14; Figure 1B, entire cohort). The double-negative expansion improved significantly only in the patients with effective B cell depletion (Figure 1A, and data not shown) (P ⫽ 0.05 versus baseline). Baseline plasmablast expansion also improved significantly after treatment and immune reconstitution (a decrease from 18.5% to 3.9% for the group overall and to 2.4% for the patients with effective B cell depletion; P ⫽ 0.009) (Figure 2A, patient 14; Figure 2B, entire cohort). In contrast, pre-GC expansion did not improve (Figure 2A, patient 18). Because plasmablasts (CD20⫺) are not directly targeted by rituximab, we were interested in the time course of disappearance of plasmablasts from the peripheral blood after treatment. We were surprised to observe a significant improvement in peripheral plasmablast expansion early after administration of rituximab in select patients. Patient 11 (in the intermediate-dose group) had a high percentage of peripheral blood plasmablasts at baseline (40%; absolute number 3.59 cells/ ␮l) and a significantly decreased, although still detectable, population at 2 months postinfusion (14%; absolute number 0.22 cells/␮l; Figure 2C). A similar pattern was observed for patient 14 (data not shown). Variability in B cell depletion. B cell depletion was highly variable in this cohort (18,20). In 6 of 17 patients, peripheral CD19⫹ lymphocytes were not depleted to ⬍1% of the total PBLs or ⬍5 cells/␮l; these 6 patients included 3 of 7 in the intermediate-dose group and 1 of 4 in the high-dose group. We previously reported that serum rituximab levels and Fc␥ receptor IIIa genotype are important determinants of B cell depletion (20). Studies in mice suggest that lymphocytes may be more resistant to depletion with monoclonal antibodies in autoimmune compared with nonautoimmune strains (30,31). In order to determine whether incomplete depletion resulted from resistance to rituximab in specific populations, we examined peripheral B cell subsets at baseline and at various times after treatment. On average, there was still a significant reduction (mean ⫾ SD 64 ⫾ 32%) in B cells at the point of maximal depletion (1–2 months posttreatment) in the 6


Figure 3. Residual B cell populations after rituximab-mediated depletion. A, B cell depletion is highly variable, with 6 of 17 patients displaying ineffective depletion (⬎5 cells/␮l remaining). To determine what B cell populations remain in these patients, peripheral blood B cells (gating on CD19⫹ B cells) were examined by flow cytometry for the expression of CD38 and IgD in SLE patient 18 after treatment (trough B cell level 2.6%) and SLE patient 6 after treatment (trough B cell level 2%), compared with a normal control. Remaining B cell populations are heterogeneous, as evidenced by division into mature B (Bm) subsets (Bm1–Bm5), as described in Patients and Methods. Patient 18 had a significant pre-GC expansion at baseline (see Figure 2A), which increased further after treatment (3 months) (rectangle gate). Patient 6 had a baseline plasmablast expansion that persisted after treatment (1 year) (circle gate). For patient 6, CD20⫹ B cells are depicted in red, and additional analysis of CD20 versus CD19 expression on total PBLs (for further identification of the PCs) is shown. B, Even after effective B cell depletion, residual B cells of a memory (M) and PC phenotype were detectable. Six months after high-dose rituximab infusion, B cells from patient 15 (⬎95% depletion; trough B cell level ⬍0.02%) were enriched by negative selection and examined for CD20, CD38, CD27, and IgD expression. Memory B cells predominated (90%), although the absolute numbers of memory cells were significantly decreased at 0.21 cells/␮l (versus 76.8 cells/␮l at baseline). See Figure 2 for other definitions.

patients with incomplete depletion. Based on their surface phenotype, residual B cells were heterogeneous and included naive (Bm1), activated naive (Bm2), pre-GC (Bm2⬘), memory (Bm5), and plasma cell phenotypes (Figure 3A, patients 18 and 6) (see Table 2 for subset definitions). Interestingly, all 3 patients with a substantial expansion of IgD⫹,CD38high circulating pre-GC cells at baseline experienced either incomplete (patients 8 and 18) or transient (patient 16) B cell depletion with a relative increase of these cells after treatment: for patient 18, 34%, 45%, and 46% of the total peripheral B

3585

cells at baseline, 3 months, and 12 months, respectively (Figures 2A and 3A); for patient 8, 35%, 70%, and 80%, respectively (data not shown); for patient 16, 25% at baseline. This suggests either that pre-GC cells are relatively resistant to depletion with rituximab, or that expansion of these cells at baseline represents a biologic marker for a subset of SLE patients who may be less responsive to treatment. Of note, these cells do express CD20 (Figure 2A), although it is still possible that they are resistant to depletion with rituximab for other reasons. Four of the 6 patients with incomplete B cell depletion (patients 3, 6, 7, and 8) had other markers of peripheral B cell activation at baseline, including an expansion of CD19low,CD38high plasma cells. This is depicted for patient 6 in Figure 3A, where plasmablasts represent 51% of the peripheral B cells at baseline and 27% posttreatment. Plasma cell expansion at baseline was not necessarily predictive of incomplete depletion, because it was common in this cohort of patients with active SLE and was also present, to varying degrees, in 7 of 11 good depletors (Figure 2B). No distinctive B cell abnormalities were identified in the remaining incomplete depletor (patient 17 in the intermediate-dose group), but this patient (along with patient 3) was unique in having no detectable serum rituximab at any time point. Such nonresponders in whom the pharmacokinetics of rituximab are rapid have been described in the lymphoma literature but remain largely unexplained (32). Residual B cell populations in patients with effective B cell depletion. Even in patients with effective B cell depletion to ⬍5 cells/␮l (95% to ⬎99% reduction), residual CD19⫹ B cells could be detected in the peripheral blood at the point of maximal depletion, prior to immune reconstitution (Figure 3B). The remaining B cells are predominantly of a switched memory phenotype (CD20⫹,CD38low, CD27⫹,IgD⫺), with a smaller percentage representing CD20⫺,CD38high plasma cells (Figure 3B). Small numbers of residual memory B cells and plasmablasts were also detected in patient 11 in the intermediate-dose group (Figure 2C) and in patient 14 in the high-dose group (data not shown), even at the point of maximal B cell depletion. It remains unclear why CD20⫹ memory B cells were still present, although the question is raised of whether B cells in peripheral lymphoid tissue may be resistant to full depletion and then experience enhanced recirculation as the peripheral blood lymphoid pool is depleted (33). Effect of B cell depletion on autoreactive B cells. Given the positive clinical response in patients with

3586

ANOLIK ET AL

Figure 4. Variable improvement of serum autoantibody titers and autoreactive memory B cells after rituximab treatment. A, Serum anti–double-stranded DNA (anti-dsDNA) titers were measured by enzyme-linked immunosorbent assay at baseline and at various time points after treatment with rituximab. Anti-dsDNA levels and Systemic Lupus Activity Measure (SLAM) scores are shown for the 8 patients with elevated titers at baseline, with effective B cell depletion (right axis) shown as a percentage of baseline levels. Improvement in autoantibody levels was observed in patients 2, 5, 9, and 14. Absolute (Abs) B cell numbers (cells/␮l) are shown on the left axis. Normal B cell counts (range 50–375 cells/␮l) are shown as dotted lines, with normalization of lymphopenia in select patients. B, Specific autoreactive VH4.34 memory B cells (CD27⫹,IgD⫺) detected by flow cytometry using the 9G4 antiidiotype, as depicted in the histograms. VH4.34 memory B cell representation is significantly higher in the SLE patients at baseline compared with normal controls and SLE patients after rituximab treatment. P values were determined by Student’s t-test. See Figure 2 for other definitions.

effective B cell depletion (18) and improvement in B cell abnormalities after rituximab treatment, we next asked whether treatment alters autoreactive B cells and autoantibodies. For this analysis, we focused on patients with effective B cell depletion and elevated levels of antidsDNA at baseline (n ⫽ 8). Surprisingly, serum antidsDNA antibody titers did not change significantly by 1 year posttreatment (mean ⫾ SD decrease of 85 ⫾ 55% of pretreatment values [range 17–164%]; P ⫽ 0.3 versus baseline). On closer inspection, 4 of 8 patients (patients 2, 5, 9, and 14) with effective B cell depletion did have a decrease in serum anti-dsDNA at 1 year posttreatment (Figure 4A) (n ⫽ 4 serologic responders; decrease to 37 ⫾ 21% [range 17–59%]; P ⫽ 0.005 versus baseline),

which became more significant with prolonged followup (decrease to 11 ⫾ 10% at 24–36 months [range 2.4– 25%]; P ⫽ 0.0002 versus baseline) (geometric mean titer, 482 at baseline and 38 at 24–36 months). Interestingly, 3 of the 4 serologic responders had a baseline absolute total B lymphopenia (⬍50 cells/␮l), which resolved after immune reconstitution (Figure 4A). However, the 4 patients with elevated anti-dsDNA at baseline and effective B cell depletion who did not have a serologic response by 1 year (patients 1, 4, 12, and 15) still had no serologic improvement with prolonged followup, and none had this robust normalization of B cell counts. This variability in anti-dsDNA response contrasts with anti-tetanus and anti-pneumococcal IgG lev-


els, which we reported were maintained in all patients at 6 months (18). As previously noted (18), in all patients SLE disease activity improved with effective B cell depletion, despite the absence of serologic improvement in some (Figure 4A). We examined VH4.34 IgG autoantibody titers by ELISA at baseline, 1 year, and during prolonged followup and found that these antibodies followed a pattern similar to that for anti-dsDNA (data not shown). An advantage of using VH4.34 as a marker of autoreactivity is that a specific antiidiotype is available for detecting B cells of this specificity in different peripheral B cell subsets, using flow cytometry. We previously reported that in normal controls VH4.34 B cells are highly censored in the GCs and are, as a result, significantly underrepresented in the memory compartment (mean ⫾ SD 1.3 ⫾ 0.3%) (23). In contrast, we observed that memory VH4.34 B cells were significantly increased in 6 SLE patients evaluated before rituximab (16.2 ⫾ 11.9%) compared with normal controls (P ⫽ 0.03), indicative of a failure of B cell tolerance. After treatment, the levels of SLE VH4.34 memory B cells returned to levels similar to those found in normal controls (1.92 ⫾ 0.7% for depletors; n ⫽ 9) (Figure 4B). DISCUSSION The results of our study demonstrate that CD20targeted B cell depletion in the treatment of SLE effectively normalizes the significant disturbances in peripheral B lymphocyte homeostasis that are characteristic of active disease, including naive lymphopenia, expansion of a population of IgD/CD27 double-negative cells, the presence of plasma cell precursors, and expansion of autoreactive memory B cell populations (14). Consistent with this normalization of B cell homeostasis, we recently reported that the subset of patients with effective B cell depletion had a significant clinical improvement as measured by the SLAM (18), despite the fact that the majority of patients had persistently elevated serum autoantibody levels. These results provide support for the notion, initially advanced in murine lupus, that B lymphocytes play a central role in human SLE independent of autoantibody production, and highlight the benefit of further exploring B cell depletion as a treatment of active SLE. One surprising finding in our study was the remarkable variability in B cell depletion achieved. Multiple factors likely underlie this variability, including differences in rituximab pharmacokinetics and Fc receptor polymorphisms that directly impact effector mecha-

3587

nisms of B cell killing mediated by the anti-CD20 monoclonal antibody (20,34). Moreover, this was a dose-escalation study, with only 4 subjects receiving high-dose rituximab; thus, the frequency of depletion failure in patients receiving full-dose therapy remains to be determined. However, our results suggest that disease activity and B cell activation may also be associated with impaired B cell depletion. Thus, SLE patients with circulating CD38high B cells (IgD⫹ GC founder cells and IgD⫺ plasma cell precursors) were more likely to have incomplete B cell depletion. There is precedent for this finding in that B cells from autoimmune mice appear to be more resistant to depletion with monoclonal antibodies than are their nonautoimmune counterparts (30). Thus, lupus B cells might be globally resistant to depletion because of intrinsic abnormalities or anomalous costimulatory signals (delivered by cytokines or through cognate interactions) that could favor B cell survival over apoptosis (35). We also recently reported that B cell activation is associated with down-modulation of surface CD20 expression via movement into lipid rafts and endocytosis, providing a potential mechanistic explanation for resistance to B cell depletion (36). Alternatively, discrete B cell subsets that are preferentially expanded in active SLE, such as CD38high GC founder cells, might be less susceptible to rituximab, thereby accounting for the ineffective depletion observed in some patients. In turn, incomplete depletion of GC founder cells could reflect either intrinsic resistance of these cells and/or ongoing exuberant GC reactions that may be central to the pathogenesis of SLE, as we recently suggested (37). Therefore, our own work indicates that the expansion of autoreactive VH4.34 IgG memory cells in SLE represents defective censoring of these cells in the GC in a disease-specific manner not shared by other autoimmune conditions, such as rheumatoid arthritis (ref. 37, and Anolik J, et al: unpublished observations). An association between incomplete B cell depletion and overactive GC reactions in SLE is also suggested by the fact that residual switched memory B cells (the presumed product of GC reactions) were still detectable in the peripheral blood of SLE patients who had effective B cell depletion. Whether B cells in the peripheral lymphoid tissue of SLE patients are resistant to full depletion with rituximab needs to be confirmed—a feasible goal for future clinical trials through tonsil biopsy of select patients. Regardless of the mechanisms responsible for incomplete B cell depletion in our cohort, the detection of even small numbers of residual peripheral blood memory and plasma cells after rituximab therapy raises

3588

important questions about the degree of B cell depletion necessary for clinical and/or serologic response or cure. The correlation between clinical response and effective B cell depletion (arbitrarily defined as depletion of B cells to ⬍1% of the total peripheral blood lymphocytes [absolute cell count ⬍5 cells/␮l]) suggests that complete B cell depletion is not necessary for clinical response. However, it is possible that more complete depletion or repeated depletion with multiple courses of rituximab will be necessary for a lasting clinical response and full restoration of B cell tolerance in the majority of patients. Thus, complete depletion of pathogenic autoreactive B cells for a prolonged period of time may be necessary (although perhaps not sufficient) to normalize abnormalities in other immune compartments that may be critical to disease pathogenesis, including T cells and dendritic cells (4,38–41). In turn, this normalization in other cell compartments may be critical to preventing the reemergence of an autoreactive B cell repertoire upon immune reconstitution (42–45). Ideally, B cell depletion provides the immune system with a second chance for proper regulation of emerging autoreactive B lymphocytes and establishment of a normal B cell repertoire (i.e., restoration of tolerance). However, the precise nature and appropriate measures of B cell tolerance in human SLE remain elusive, although critically important to define. Thus, as highlighted by the variability in serum autoantibody response in our study, measurement of autoantibodies represents only an indirect marker that could also reflect the persistence of plasma cells with a heterogeneous lifespan. We suggest that determining the fate of autoreactive VH4.34 B cells represents an additional and powerful biomarker of tolerance in general and of GC censoring in particular. Therefore, normalization of autoreactive VH4.34 memory B cells in select patients treated with rituximab may reflect restoration of a B cell tolerance checkpoint at the level of the GC, although proof of this hypothesis will require examination of peripheral lymphoid tissue. Moreover, how reproducible this finding is in different SLE patients and the immunologic basis of response variability are important areas for future study. The observed variability in autoantibody response also raises important questions regarding autoreactive plasma cell biology in SLE. Autoreactive B cells could be successfully and fully eliminated by rituximab and not reemerge upon immune reconstitution, yet anti-dsDNA levels (and VH4.34 antibodies) remain elevated because of the presence of long-lived autoreactive plasma cells that continue to produce autoantibody.

ANOLIK ET AL

Alternatively, precursor B cells (memory B cells in particular, based on our results) may be incompletely depleted and provide a reservoir for the continuous production of autoreactive plasma cells. Distinguishing between these 2 possibilities is quite difficult but has important implications for the pathogenesis and treatment of human SLE and the origin of autoimmune memory. Indeed, controversy has surrounded the question of whether autoimmune memory in SLE is attributable to long-lived autoreactive plasma cells and/or continuous stimulation of autoreactive memory B cells, through either self antigen or polyclonal activation (46,47). Although the latter process likely plays an important role, there is increasing experimental evidence that a fraction of plasmablasts are long-lived, are important in maintaining humoral antibody memory, and can contribute to autoimmune disease. As reported by Slifka et al (48), in mice noncycling bone marrow– derived plasmablasts can be transferred to antigen-naive mice and survive and secrete antibody for more than a year in the absence of detectable memory B cells (48). In autoimmune NZB/NZW mice, although the continuous production of high numbers of short-lived plasma cells is important to hypergammaglobulinemia and the autoimmune process, a fraction of the autoreactive plasmablasts are noncycling and long-lived (49,50). Moreover, these long-lived plasma cells reside not only in the bone marrow, as conventionally thought, but also in additional survival niches (e.g., the kidneys and possibly other inflamed tissues) and may be particularly resistant to treatment interventions (47,50). In human SLE, the persistence of some autoantibodies even during quiescent disease or following immunosuppression has been taken to reflect the presence of long-lived plasma cells. We had originally surmised that anti-dsDNA is likely produced by short-lived plasma cells, because the levels of these autoantibodies can be dramatically reduced by treatment with high-dose steroids with or without immunosuppressive drugs, whereas the levels of total IgG and other autoantibodies such as anti-Sm are, in contrast, unchanged. The existence of short-lived plasma cell populations in SLE is supported by our data demonstrating a significant decrease in peripheral plasmablast expansion immediately after rituximab treatment in select patients. Although it is possible that the peripheral blood plasmablasts did not die off but rather homed to the bone marrow, other lymphoid tissue, or inflamed tissue, this would be less consistent with the serum autoantibody normalization in these patients. Regardless, the fact that rituximab treatment significantly decreased plasma cell expansion in


this cohort highlights the fact that B cell depletion can alter plasma cell homeostasis in SLE. In contrast, the persistence of elevated antidsDNA and VH4.34 autoantibody titers in the majority of our patients despite normalization of autoreactive memory B cell numbers in some suggests that the lifespan of autoreactive plasmablasts, even those conventionally thought of as short-lived, may vary in different patients. In fact, the lifespan of autoreactive plasmablasts may be heterogeneous even in individual patients. This possibility is supported by the kinetics of the anti-dsDNA decrease in serologic responders, with antidsDNA titers decreasing immediately after rituximab treatment but not normalizing for well over 12 months. As described for NZB/NZW mice, the survival of autoreactive plasmablasts in human SLE similarly may be dependent on microenvironment. Whether plasmablast survival niches are provided by the bone marrow or other sites (such as inflamed target tissue) and how steroids or other treatment agents might alter the microenvironment are important unresolved questions. In conclusion, this study is the first to show evidence that in SLE, B cell depletion therapy with rituximab dramatically improves abnormalities in B cell homeostasis, with a decreased proportion of autoreactive memory B cells after treatment. The persistence of serum autoantibodies in most patients is noteworthy and raises important questions regarding the relative contribution of long-lived autoreactive plasma cells versus memory B cells to the autoimmune process in SLE. It seems reasonable that for maximal clinical efficacy and induction of long-term remissions with rituximab, full reestablishment of B cell tolerance with elimination of autoreactive memory and plasma cell populations will be necessary. Whether this goal can be consistently achieved using repeated cycles of rituximab and/or combination therapy is an important area of future investigation. ACKNOWLEDGMENTS We are grateful to Debbie Campbell for assistance with patient recruitment. We acknowledge the support of Genentech (South San Francisco, CA) and IDEC Pharmaceuticals (San Diego, CA) for the clinical study.

REFERENCES 1. Chan OT, Hannum LG, Haberman AM, Madaio MP, Shlomchik MJ. A novel mouse with B cells but lacking serum antibody reveals an antibody-independent role for B cells in murine lupus. J Exp Med 1999;189:1639–48.

3589

2. Shlomchik MJ, Madaio MP, Ni D, Trounstein M, Huszar D. The role of B cells in lpr/lpr-induced autoimmunity. J Exp Med 1994;180:1295–306. 3. Chan OT, Madaio MP, Shlomchik MJ. The central and multiple roles of B cells in lupus pathogenesis. Immunol Rev 1999;169: 107–21. 4. Harris D, Haynes L, Sayles P, Duso DK, Eaton SM, Lepak NM, et al. Reciprocal regulation of polarized cytokine production by effector B and T cells. Nat Immunol 2000;1:475–81. 5. Wakeland E, Liu K, Graham R, Behrens T. Delineating the genetic basis of systemic lupus erythematosus. Immunity 2001;15: 397–408. 6. Mohan C. Murine lupus genetics: lessons learned. Curr Opin Rheumatol 2001;13:352–60. 7. Qian Y, Wang H, Clarke S. Impaired clearance of apoptotic cells induces the activation of autoreactive anti-Sm marginal zone and B-1 B cells. J Immunol 2004;172:625–35. 8. Folzenlogen D, Hofer MF, Leung DY, Freed JH, Newell MK. Analysis of CD80 and CD86 expression on peripheral blood B lymphocytes reveals increased expression of CD86 in lupus patients. Clin Immunol Immunopathol 1997;83:199–204. 9. Desai-Mehta A, Lu L, Ramsey-Goldman R, Datta SK. Hyperexpression of CD40 ligand by B and T cells in human lupus and its role in pathogenic autoantibody production. J Clin Invest 1996;97: 2063–73. 10. Avery D, Kalled S, Ellyard J, Ambrose C, Bixler SA, Thien M, et al. BAFF selectively enhances the survival of plasmablasts generated from human memory B cells. J Clin Invest 2003;112:286–97. 11. Zhang J, Roschke V, Baker KP, Wang Z, Alarcon GS, Fessler BJ, et al. Cutting edge: a role for B lymphocyte stimulator in systemic lupus erythematosus. J Immunol 2001;166:6–10. 12. Odendahl M, Jacobi A, Hansen A, Feist T, Hiepe F, Burmester GR, et al. Disturbed peripheral B lymphocyte homeostasis in systemic lupus erythematosus. J Immunol 2000;165:5970–9. 13. Arce E, Jackson DG, Gill MA, Bennett LB, Banchereau J, Pascual V. Increased frequency of pre-germinal center B cells and plasma cell precursors in the blood of children with systemic lupus erythematosus. J Immunol 2001;167:2361–9. 14. Cappione A, Pugh-Bernard A, Anolik JH, Sanz I. Lupus IgG VH4.34 antibodies bind to a 220-kDa glycoform of CD45/B220 on the surface of human B lymphocytes. J Immunol 2004;172: 4298–307. 15. Anolik JH, Campbell D, Ritchlin C, Holyst M, Rosenfeld S, Varnis C, et al. B lymphocyte depletion as a novel treatment for systemic lupus erythematosus (SLE): phase I/II trial of rituximab (Rituxan®) in SLE [abstract]. Arthritis Rheum 2001;44 Suppl 9:S387. 16. Leandro M, Edwards JC, Cambridge G, Ehrenstein MR, Isenberg DA. An open study of B lymphocyte depletion in systemic lupus erythematosus. Arthritis Rheum 2002;46:2673–7. 17. Edwards J, Cambridge G. Sustained improvement in rheumatoid arthritis following a protocol designed to deplete B lymphocytes. Rheumatology 2001;40:205–11. 18. Looney RJ, Anolik JH, Campbell D, Felger RE, Young F, Arend LJ, et al. B cell depletion as a novel treatment for systemic lupus erythematosus. Arthritis Rheum 2004;50:2580–9. 19. Anolik J, Sanz I, Looney RJ. B cell depletion therapy in systemic lupus erythematosus. Curr Rheumatol Rep 2003;5:350–6. 20. Anolik JH, Campbell D, Felgar RE, Young F, Sanz I, Rosenblatt J, et al. The relationship of Fc␥RIIIa genotype to degree of B cell depletion by rituximab in the treatment of systemic lupus erythematosus. Arthritis Rheum 2003;48:455–9. 21. Tan E, Cohen A, Fries J, Masi A, McShane D, Rothfield N. The 1982 revised criteria for the classification of systemic lupus erythematosus. Arthritis Rheum 1982;25:1271–7. 22. Liang MH, Socher SA, Larson MG, Schur PH. Reliability and validity of six systems for the clinical assessment of disease activity

3590

23.

24. 25. 26.

27.

28. 29. 30.

31.

32.

33. 34. 35. 36.

in systemic lupus erythematosus. Arthritis Rheum 1989;32: 1107–18. Pugh-Bernard AE, Silverman GJ, Cappione AJ, Villano ME, Ryan DH, Insel RA, et al. Regulation of inherently autoreactive VH4-34 B cells in the maintenance of human B cell tolerance. J Clin Invest 2001;108:1061–70. Maurer D, Fischer GF, Fae I, Majdic O, Stuhlmeier K, Von Jeney N, et al. IgM and IgG but not cytokine secretion is restricted to the CD27⫹ B lymphocyte subset. J Immunol 1992;148:3700–5. Pascual V, Liu YJ, Magalski A, de Bouteiller O, Banchereau J, Capra JD. Analysis of somatic mutation in five B cell subsets of human tonsil. J Exp Med 1994;180:329–39. Bohnhorst JO, Bjorgan MB, Thoen JE, Natvig JB, Thompson KM. Bm1-Bm5 classification of peripheral blood B cells reveals circulating germinal center founder cells in healthy individuals and disturbance in the B cell subpopulations in patients with primary Sjo ¨gren’s syndrome. J Immunol 2001;167:3610–8. Harada Y, Kawano M, Huang N, Mahmoud MS, Lisukov IA, Mihara K, et al. Identification of early plasma cells in peripheral blood and their clinical significance. Br J Haematol 1996;92: 184–91. Anolik JH, Cappione A, Looney RJ, Sanz I, Young F. Abnormalities in B lymphocyte signaling in human SLE. Arthritis Rheum 2002;46:S415. Huang W, Sinha J, Newman J, Reddy B, Budnai L, Furie R, et al. The effect of anti-CD40 ligand antibody on B cells in human systemic lupus erythematosus. Arthritis Rheum 2002;46:1554–62. Cerny A, Starobinski M, Hugin AW, Sutter S, Zinkernagel RM, Isui S. Treatment with high doses of anti-IgM prevents, but with lower doses accelerates autoimmune disease in (NSW x BXSB)F1 hybrid mice. J Immunol 1987;138:4222–8. Connolly K, Roubinian J, Wofsy D. Development of murine lupus in CD4-depleted NZB/NZW mice: sustained inhibition of residual CD4⫹ T cells is required to suppress autoimmunity. J Immunol 1992;149:3083–8. Grillo-Lopez A, White C, Varns C, Shen D, Wei A, McClure A, et al. Overview of the clinical development of rituximab: first monoclonal antibody approved for the treatment of lymphoma. Semin Oncol 1999;26:66–73. Schroder C, Azimzadeh A, Wu G, Price J, Atkinson J, Pierson R. Anti-CD20 treatment depletes B cells in blood and lymphatic tissue of cynomolgus monkeys. Transpl Immunol 2003;12:19–28. Johnson P, Glennie M. The mechanisms of action of rituximab in the elimination of tumor cells. Semin Oncol 2003;30:3–8. Silverman G, Weisman S. Rituximab therapy and autoimmune disorders: prospects for anti–B cell therapy. Arthritis Rheum 2003;48:1484–92. Anolik JH, Looney RJ, Bottaro A, Sanz I, Young F. Down-

ANOLIK ET AL

37. 38. 39. 40. 41.

42. 43. 44. 45. 46. 47. 48. 49. 50.

51.

regulation of CD20 on B cells upon CD40 activation. Eur J Immunol 2003;33:2398–409. Cappione A, Anolik J, Barnard J, Pugh-Bernard A, Dutcher P, Sanz I. Defective germinal center censoring of autoreactive B cells in human SLE. Arthritis Rheum 2003;48:S193. Mamula MJ, Fatenejad S, Craft J. B cells process and present lupus autoantigens that initiate autoimmune T cell responses. J Immunol 1994;152:1453–61. Linton PJ, Harbertson J, Bradley LM. A critical role for B cells in the development of memory CD4 cells. J Immunol 2000;165: 5558–65. Skok J, Poudrier J, Gray D. Dendritic cell-derived IL-12 promotes B cell induction of Th2 differentiation: a feedback regulation of Th1 development. J Immunol 1999;163:4284–91. Moulin V, Andris F, Thielemans K, Maliszewski C, Urbain J, Moser M. B lymphocytes regulate dendritic cell (DC) function in vivo: increased interleukin 12 production by DCs from B cell deficient mice results in T helper cell type 1 deviation. J Exp Med 2000;192:475–82. Goodnow CC, Cyster JG, Hartley SB, Bell SE, Cooke MP, Healy JI, et al. Self-tolerance checkpoints in B lymphocyte development. Adv Immunol 1995;59:279–368. Meffre E, Schaefer A, Wardemann H, Wilson P, Davis E, Nussenzweig M. Surrogate light chain expressing human peripheral B cells produce self-reactive antibodies. J Exp Med 2004;199:145–50. Wardemann H, Yurasov S, Schaefer A, Young J, Meffre E, Nussenzweig M. Predominant autoantibody production by early human B cell precursors. Science 2003;301:1374–7. Ueda Y, Yang K, Foster S, Kondo M, Kelsoe G. Inflammation controls B lymphopoiesis by regulating chemokine CXCL12 expression. J Exp Med 2004;199:47–58. Bernasconi N, Traggiai E, Lanzavecchia A. Maintenance of serologic memory by polyclonal activation of human memory B cells. Science 2002;298:2199–202. Manz R, Arce S, Cassese G, Hauser A, Hiepe F, Radbruch A. Humoral immunity and long-lived plasma cells. Curr Opin Immunol 2002;14:517–21. Slifka M, Antia R, Whitmire J, Ahmed R. Humoral immunity due to long-lived plasma cells. Immunity 1998;8:363–72. Cassese G, Lindenau S, de Boer B, Arce S, Hauser A, Riemekasten G, et al. Inflamed kidneys of NZB/W mice are a major site for the homeostasis of plasma cells. Eur J Immunol 2001;31:2726–32. Hoyer B, Moser K, Hauser A, Peddinghaus A, Voight C, Eilat D, et al. Short-lived plasmablasts and long-lived plasma cells contribute to chronic humoral autoimmunity in NZB/W mice. J Exp Med 2004;199:1577–84. Carsetti R, Rosado MM, Wardmann H. Peripheral development of B cells in mouse and man. Immunol Rev 2004;197:179–91.