Patients

Data were obtained from outpatients attending a lupus clinic that was started in January 2010 as a joint initiative of nephrologists and rheumatologists at Münster University Hospital. All patients underwent standard laboratory tests allowing an assessment of disease activity as well as the safety of their current therapeutic regimen. These tests include all routine diagnostic procedures and established biomarkers, allowing an assessment of disease activity, and the patient stratification needed to decide on the most suitable therapeutic approach. Although no trial was planned, a characteristic pattern of B cell subsets became apparent in patients on MMF. A retrospective analysis of clinical, serological and cellular parameters was performed to compare patients taking MMF with patients taking AZA, and patients not taking immunosuppressive treatment. Ethical approval and informed consent for monitoring as well as retrospective data analysis were waived by the local ethics committee. All patients attending the clinic within its first year were included in this analysis, provided they had been taking their current medication for at least six months. Patients treated with rituximab at any time were excluded, since it causes a sustained alteration of B cell subsets.

Flow cytometric analysis of peripheral blood lymphocyte subsets

Flow cytometric analysis of peripheral blood mononuclear cells (PBMC) is a procedure performed routinely in addition to differential blood count in patients undergoing immunosuppressive therapy. From 3 ml of heparinized blood PBMCs were isolated by density gradient centrifugation using Ficoll-Paque™ Plus from GE Healthcare (Munich, Germany), were washed in PBS/0.5% BSA (Sigma-Adlrich, St. Gallen, Switzerland) and stained immediately with fluorochrome-labeled monoclonal antibodies to a panel of different surface antigens to discriminate B and T cell subsets (see Additional file 1). All samples were processed and analyzed within 6 hours after collection to ensure viability of all cell subsets. To exclude dead cells a final concentration of 220 nM 4´,6-diamidino-2-phenylindole (DAPI) (Invitrogen, Carlsbad, CA, USA) was used. A FACS Canto-II and FACS Diva Software (Becton Dickinson, (BD), San Jose, CA, USA) were used for 12-parameter (8-color) flow cytometric analysis. One million events were recorded for B cell and 500,000 events for T cell analysis. Results were analyzed using FlowJo (Treestar, Ashland, OR, USA).

Lymphocyte counts were recorded and absolute numbers were calculated using the frequencies of T and B cells based on the lymphocyte gate and the lymphocyte count. Differential blood counts and all other lab values including autoantibody titers were determined in the central laboratory using accredited diagnostic procedures [35].

Isolation of B cell subsets for functional analysis

Naïve and memory B cells from blood donors (leukocyte filters) were isolated in a multistep procedure. The use of leukocyte filters from healthy blood donors for in vitro assays was approved by the local ethics committee. First the filter content was incubated with RosetteSep B Cell Enrichment Cocktail (STEMCELL Technologies SARL, Grenoble, France) according to the manufacturer's' instructions, and density gradient centrifugation was performed. Isolated B cells were then labeled with anti-CD27-magnetic beads (Miltenyi Biotec, Bergisch Gladbach, Germany) and CD27⁺ memory B cells were positively selected using a magnetic column and the Miltenyi Biotec protocol. Subsequently the remaining B cells were labeled with biotin-labeled anti-IgD antibody (IA6-2, BD, Bioscience, Franklin Lakes, NJ, USA). After washing cells twice with PBS/0.5%BSA, spreptavidin-labeled magnetic beads (Miltenyi Biotec) were used for positive selection of CD27^-IgD⁺ naïve B cells on a second column.

Flow cytometric analysis after magnetic activated cell sorting (MACS) confirmed that CD3⁺ T cells were depleted completely from the CD27⁺ subset.

Proliferation assay

Naïve and memory B cells were obtained as described above and subsequently stained with carboxyfluorescein-succinimidyl-ester (CFSE, Invitrogen) according to the manufacturers' instructions. CFSE-labeled cells were then incubated at a concentration of 10⁵/ml in RPMI 1640 (GIBCO, Invitrogen) supplemented with 10% v/v FCS, penicillin (100 U/ml) and streptomycin (100 mg/ml) (all Invitrogen) on a 96-well round-bottom plate (Greiner Bio-One, Kremsmuenster, Austria) with CpG oligodeoxynucleotide 2006, sequence: 5'-TsCsg sTsCsg sTsTsT sTsgsT sCsgsT sTsTsT sgsTsC sgsTsT-3' 2.5 µg/ml (TIB MolBiol, Berlin, Germany) or 0.5 µg/ml of a monoclonal anti-CD40 antibody (MAB89 (Beckmann Coulter, Eurocenter S.A., Nyon, Switzerland) and 50 ng/ml of IL-21 (GIBCO, Invitrogen) with or without 5 µM of MPA (Sigma-Aldrich). Four days later, cells were harvested, washed with PBS, and stained with monoclonal antibodies (see Additional file 1) in a 96-well v-bottom plate (Brand, Wertheim, Germany). Proliferation assays were performed in doublets using peripheral B cell subsets from four healthy blood donors. Flow cytometric analysis and data processing were performed as described.

Determination of STAT3 phosphorylation by phospho-flow cytometry

PBMC were isolated by density gradient centrifugation and incubated with 5 µM MPA, or without MPA, for 24 hours in RPMI1640 supplemented with 10%v/v FCS, penicillin (100 U/ml) and streptomycin (100 mg/ml). Subsequently cells were stimulated with IL-21 (50 ng/ml) for 15 minutes at 37°C and fixed in 1.5% formaldehyde (ROCKLAND Gilbertsville, PA, USA) for 10 minutes at room temperature. Afterwards PBMC were washed in PBS/0.5% BSA and stained with fluorescence-labeled monoclonal antibodies to a panel of different surface antigens to allow the discrimination of B and T cell subsets (see Additional file 1). After another washing step, cells were incubated with methanol (ice-cooled, 100%, Riedel-de-Haen AG, Seelze, Germany) for 10 minutes at 4°C followed by two more wash steps. Next intracellular labeling of pSTAT3 was performed incubating cells with a monoclonal antibody (see Additional file 1) for 30 minutes at 4°C. All samples were acquired immediately after two more wash steps using a FACS Canto-II equipped with FACS Diva Software (BD) and data were subsequently analyzed using FlowJo software (Treestar). One million events were recorded.

Statistical analysis

Frequencies of lymphocyte subsets analyzed ex vivo or in vitro were calculated using FlowJo software (TreeStar). Differences in frequencies or numbers of certain cell subsets were determined using the Kruskal-Wallis test and Dunn's multiple comparison test, since the majority of data were not normally distributed. Except for age (presented as mean ± SD), median values with the range are shown. The Chi-square test was used to determine if organ involvement, co-medications or disease flares were significantly over-, or under-represented in any of the patient cohorts. The Wilcoxon matched-pairs signed rank test was performed to compare in vitro cell survival with and without MPA. P-values < 0.05 were considered statistically significant. Data were analyzed using GraphPad Prism5 (GraphPad, San Diego, CA, USA).

Clinical aspects

Although AZA- and MMF-treated patients had not been randomized for important parameters such as disease activity, prednisone-equivalent dose, disease manifestations or autoantibody profile, there were no significant differences in the occurrence of any parameter analyzed as shown in Table 1.

Disease flares requiring intensification of the current treatment regimen by increasing the prednisone-equivalent dose to > 50 mg/day and/or the dose, or type of immunosuppressant, were observed more often in patients taking AZA (47%) compared to MMF-treated patients (23%). Organ involvement observed during flares is shown in Table 1.

Effect of AZA and MMF on peripheral blood lymphocytes

Differential blood counts and flow cytometric data were analyzed to estimate and compare the effect of AZA and MMF on cellular parameters. A summary of cell counts is shown in Table 2. Although patients without immunosuppressive treatment had higher T cell numbers than patients treated with MMF or AZA, no significant differences were observed. T cell subset analysis did not reveal significant differences between AZA- and MMF-treated patients and patients without immunosuppressive therapy. However, when B cell subsets were compared, a couple of significant differences were identified. Performance and results of the B cell subset analysis are shown in Figures 1, 2, 3.

	AZA	MMF	No IS
Hemoglobin, mmol/l	7.9 (5.5-10.4)	8.2 (5.4-10.7)	8.2 (4.5-10.7)
Platelets/µl	238 (48-459)	252 (136-404)	223 (96-379)
Leukocytes	4855 (2,060-10,160)	5970 (2,640-15,550)	5225 (2,080-9,050)
Lymphocytes/µl	987 (200-3,790)	1080 (360-3,010)	1330 (300-2,910)
B cells %	4.5 (0.6-31.3)* §§§	9.4 (0.6-40.1)*	11.8 (3.6-28.9)§§§
B cells/µl	44(3-662)** §§§	113(4-462)**	119 (40-460)§§§
T % of L	76 (28-93)*	63 (24-85)*	64 (42,2-84,5)
T/µl	584 (180-3,286)	640 (89-2,468)	807 (162-2,104)
CD4+ %	73 (39-82)	71 (41-92)	69 (33-92)
CD4+/µl	430 (85-1,709)	466 (61-2,021)	518 (106-1,514)
CD8+ %	19 (7-46)	22 (3-48)	22 (2-60)
CD8+/µl	127 (25-1,410)	144 (12-696)	207 (19-604)
CD4-/CD8- %	6 (2-18)	6 (2-11)	5 (2-20)
CD4-/CD8- /µl	41 (8-136)	41 (2-139)	46 (4-137)

Results are presented as median (range). Statistically significant differences detected between cell counts of patients taking azathioprine (AZA) and mycophenolate mofetil (MMF) (*P < 0.05, **P < 0.01, ***P < 0.001) and AZA compared to patients not taking immunosuppressive therapy ^§P < 0.05, ^§§P < 0.01, ^§§§P < 0.001 (Dunn's multiple comparison test).

Although patients on AZA had significantly lower B cell counts compared to untreated patients or patients taking MMF (P = 0.0001) (Table 2), CD27^highCD38^high ASCs were found at a significantly higher frequency and similar number in the peripheral blood of these patients compared to patients without immunosuppressive therapy (data not shown). Importantly, patients taking MMF had significantly lower frequencies and numbers of ASCs in the peripheral blood compared to AZA-treated patients, or patients without immunosuppressive therapy (P < 0.0001 respectively) (data not shown).

In order to analyze if the impact of MMF on the ASC subset is restricted to recently produced HLA-DR^high plasmablasts that usually represent the vast majority of the CD27^highCD38^high ASC subset in patients with flaring lupus, peripheral HLA-DR^low ASC and HLA-DR^high ASC were determined and compared (Figure 2). Although the absolute number of HLA-DR^low ASC was significantly lower in patients on MMF (P = 0.017, Figure 2B), this deviation was moderate compared to the marked difference of HLA-DR^high ASC counts or frequencies observed between patient groups (P < 0.0001, Figure 2A). Therefore, MMF treatment was primarily associated with decreased frequencies and numbers of HLA-DR^high ASC in the peripheral blood of lupus patients.

Frequencies and numbers of CD27^-IgD⁺CD38⁺⁺ transitional B cells were significantly lower in patients taking AZA compared to MMF-treated patients, or patients without immunosuppressive therapy (P < 0.0001, Figure 3A). In addition, frequencies and numbers of CD27^-IgD⁺CD38⁺ mature naïve B cells were significantly lower in patients taking AZA compared to MMF-treated patients, or patients without immunosuppressive therapy (P = 0.0001 and P < 0.0001, respectively, Figure 3B).

In contrast, memory B cell analyses revealed no difference, or less pronounced differences, between patient groups (Figures 3C and 3D). Median frequencies of CD27^-IgD^- B cells (heterogeneous subset also containing memory B cells [36]) were comparably high in patients on MMF (16.2%, range 5.7 to 41.7%), AZA (17.7%, 7.7 to 36.1%) and patients not on immunosuppressive drugs (17.0%, 4.0 to 33.0%) (data not shown). Also median frequencies of CD27⁺IgD⁺ pre-switched memory B cells did not differ (Figure 3C). The median frequencies of CD27⁺IgD^- post-switched memory B cells were comparably high in patients on MMF and patients not on immunosuppressive drugs, whereas AZA-treated patients had significantly higher frequencies of post-switched memory B cells (P = 0.0003). This comparative increase was caused by a lack of antigen-naïve B cells observed in AZA-treated patients. However, absolute numbers of post-switched memory B cells were comparably high in all patients (Figure 3D).

Effect of MMF on IgG levels

Consistent with the results of B cell subset analysis and dsDNA antibody levels, patients taking MMF had significantly lower IgG levels compared to patients taking AZA, or patients not on immunosuppressive drugs (P = 0.0005, Figure 2C). Notably, no significant difference was found between the serum IgG levels of patients on AZA and patients not taking immunosuppressive drugs (Figure 2C). However the percentage of patients below the normal range was comparable in AZA- and MMF-treated patients (both 10.3%).

In vitro B cell proliferation and differentiation of ASCs are abolished completely by MPA

B cell proliferation assays were performed as described to investigate if the impact of MMF on B cell activation and generation of ASCs that we postulated in patients with lupus, was indirect or direct. The MPA concentration that was used in all functional assays had been chosen considering previous work, and the approximate plasma concentration reached by daily intake of 2-3 g MMF [37].

Proliferation assays of purified CD27^-IgD⁺ antigen-naïve and CD27⁺ memory B cell subsets from the peripheral blood of four healthy blood donors were performed. All individuals showed a comparable pattern of proliferation and a representative example is shown (Figure 4). In all assays MPA was able to abolish B cell proliferation and differentiation of ASCs completely, no matter if it was caused by TLR-mediated stimulation using CpG, or by CD40-antibodies in combination with IL-21, suggesting that the impact of MMF on B-cell activation and proliferation is not only related to impaired T cell help. In contrast to calcineurin inhibitors [38], MMF inhibits B cell activation and differentiation of ASCs directly.

While proliferation and formation of ASCs were abolished completely, the survival of non-stimulated B cells was only partially impaired by 5µM MPA after 4 days of culture (Figure 4). Comparing cell survival in cultures incubated with and without MPA it was reduced to a median value of 52.4% for antigen-naïve (range 34.9 to 67.8, P = 0.125) and 76.4% for CD27⁺ memory B cells (range 51.1 to 82.3, P = 0.125 (data not shown).

MPA does not affect STAT3 phosphorylation in antigen-naïve or CD27⁺ memory B cells

To rule out additional effects of MMF on mechanisms that are important for B cell proliferation and ASC formation we also analyzed STAT3 phosphorylation. STAT3 signal transduction is required to form memory B cells and ASCs [21, 22]. It is essential for humoral immune responses. Phosphorylation of STAT3 has been shown to be impaired in myeloma cells after incubation with MPA [23]. Therefore, we decided to investigate the impact of MPA on IL-21-induced STAT3 phosphorylation in antigen-naïve and CD27⁺ memory B cells from the peripheral blood of healthy individuals. As shown in Figure 5, a 24-hour pre-incubation with 5 µM MPA did not impact STAT3 phosphorylation detected after exposure of CD27^-IgD⁺ antigen-naïve or CD27⁺ memory B cells to IL-21.