Defective regulatory B-cell compartment in patients with immune thrombocytopenia

B lymphocytes participate in immune responses through production of antibodies, antigen presentation to T cells, and cytokine secretion. Animal studies suggest that naive B cells can also regulate autoimmune responses through secretion of anti-inflammatory IL-10¹  and control proinflammatory differentiation of other antigen-presenting cells (APCs).²  More recently, human B cells with regulatory functions mediated in part by IL-10 and/or through inhibitory interactions with effector T cells and monocytes have been described.^3–5  In the original report of human B-regulatory cells (Bregs), the CD19⁺CD24^hiCD38^hi human B-cell subpopulation, that included immature transitional B cells, was shown to possess regulatory activity by reducing CD4⁺ T-cell activation at least in part via IL-10 secretion.³  Interestingly, in patients with systemic lupus erythematosus (SLE), the CD19⁺CD24^hiCD38^hi subset produced less IL-10 production and had reduced suppressive activity, suggesting that altered cellular function of the Breg compartment in SLE may impact the immune effector responses in this autoimmune disease.³  Furthermore, in renal transplantation patients, increased frequency of CD19⁺CD24^hiCD38^hi was associated with positive outcome.⁶  Although human studies have yet to be performed, various mouse disease models indicate that IL-10 produced by Bregs is important for control and maintenance of regulatory T cells (Tregs),⁷  and promote their differentiation⁸  or their recruitment,⁹  thus further expanding the role of Bregs in immunoregulation.

Immune thrombocytopenia (ITP) is an autoimmune bleeding disorder because of immune destruction of platelets and insufficient platelet production. Antiplatelet autoantibodies are responsible for platelet destruction by the reticuloendothelial system and probably for inhibition of megakaryopoiesis.¹⁰  Marked reduction of memory B cells and high plasma levels of B-cell activating factor BAFF that may affect B cell compartment size and subset distribution have been reported in patients with ITP.^11–13  Furthermore, a shift toward stimulatory monocytes with enhanced FcγR-mediated phagocytic capacity may exacerbate platelet destruction, and also contribute to the autoimmune response to platelet antigens.¹⁴  A generalized altered immune regulation in ITP is further supported by presence of platelet-autoreactive T cells, cytokine imbalance,^15–19  and altered regulatory T-cell (Treg) numbers and function.^20–25  Several treatment options are available to ITP patients, including the use of thrombopoietic (TPO) agents, which have yielded overall safe and durable responses while treatment continues in patients with chronic and refractory ITP.^26–31  Interestingly, various therapies that increase platelet counts, such as rituximab, thrombopoietic agents, or treatment with high-dose dexamethasone are also associated with improved Treg function or numbers.^22,23,32 

Given that ITP pathogenesis is in part related to defective Treg, T helper, and monocyte functions and because Bregs are important for controlling effector T-cell and monocyte responses and possibly Tregs, we initiated studies to characterize the Breg compartment in ITP patients. Because we had observed an improvement in Treg activity in patients on TPO agents with increased platelet counts, we studied a cohort of patients on TPO agents with various platelet count responses. Our studies are consistent with a disturbed B-cell regulatory function in patients with ITP as with other autoimmune diseases with the potential to improve after treatment with TPO agents that increase platelet counts, similar to the findings with improved Tregs in responders to treatment.

All the studies were approved by the institutional review boards of the Weill Medical College of Cornell University and of the New York Blood Center. Patients with chronic ITP (> 1 year since diagnosis, 17 male and 19 females, n = 36) were enrolled in the study (see Tables 1,Table 2–3 for patient clinical profiles) after informed consent. Closely age-matched, healthy subjects were recruited as controls. Although not indicated, most had been refractory to previous treatments. Some patients had received rituximab several years before the study and, with only 1 exception (no. 29), were considered nonresponsive. Patients listed as being “no treatment” in Tables 1,Table 2–3 had been without ITP treatment regimen for at least 2 weeks before the study. Patients “On TPO agents” included those who were on promacta and romiplostim as well as the investigational small-molecule thrombopoietin receptor (TPO-R) agonist Shionogi S-888711. Unless indicated in Tables 1,Table 2–3, patients were not on any other ITP treatment at the time of the blood draw. Patients 30 and 35 are considered in remission since both had been without a need for treatment for more than 1 year since their last treatment: patient 30 was last treated with a TPO agent a year and a half before the study, and patient 36 had been treated with non-TPO agents 5 years before the study. Patient 29, who had a platelet count of 54 × 10⁹/L at the time of the study visit, had been treated with rituximab and had since been without a need for treatment for 9 years when she was enrolled in the study and is considered partially in remission.

B-cell phenotypic analysis was performed using whole blood (100 μL) using fluorescently conjugated antibodies against human CD19, CD24, CD38, and CD27 (all from BD Biosciences). For intracellular IL-10 expression studies, PBMCs were resuspended (2 × 10⁶ cells/mL) in media containing 10% fetal calf serum, 4mM l-glutamine, and antibiotics (all from Invitrogen Life Technologies) before stimulation with 100nM CpG ODN2006 (Invivogen) for 72 hours. PMA (50 ng/mL; Sigma-Aldrich), ionomycin (750 ng/mL; Sigma-Aldrich), and GolgiPlug (Brefeldin A; BD Biosciences) were added 5.5 hours before the end of the cultures. After B-cell surface staining, cytoplasmic IL-10 expression was analyzed using a Cytofix/Cytoperm kit (BD Biosciences) according to the manufacturer's instructions and staining with anti–human IL-10 or an isotype-matched control.

CD19⁺ and CD14⁺ cells from PBMCs were enriched using microBeads according to the manufacturer's instructions (Miltenyi Biotec). Purified (> 90%) CD19⁺ cells (2 × 10⁶/mL) were cultured for 24 hours with mega-sCD40L (50 μg/mL; Enzo Life Sciences) plus CpG ODN2006 (InVivogen; 200nM) while purified CD14⁺ (> 90%) cells were cultured alone. Washed 2 × 10⁵ CD19 cells were then added to equal numbers of CD14⁺ monocytes and cocultured for 16-20 hours. LPS (1 μg/mL) together with GolgiPlug (BD Biosciences) was added for 4 hours followed by flow cytometric analysis of surface CD14 and intracellular TNF-α expression. Inhibition of monocyte cytokine expression was calculated as 1 − (% TNF-α expressed in cocultures of activated B cells and monocytes/% TNF-α expressed in monocytes cultured alone) × 100, where 100% was considered complete inhibition and 0% no inhibition.

Statistical analyses were performed using the Mann-Whitney U test for all comparisons described in the study, and differences were considered significant at P < .05.

As a first step to analyze the regulatory B-cell compartment, we divided our chronic ITP patient cohort based on platelet counts of either below or above 50 × 10⁹/L because this platelet count is a clinical indicator of patients' disease severity. Given the role of the spleen in B-cell development, we further divided the patients for the phenotypic analysis into 2 groups consisting of splenectomized and nonsplenectomized patients. Despite comparable proportions of CD19⁺ B cells in nonsplenectomized patients with ITP (supplemental Figure 1A), the frequency of circulating CD19⁺CD24^hiCD38^hi B cells, previously shown to have regulatory function,³  was decreased in patients with low platelet counts who were off ITP treatment (n = 8, filled squares) compared with healthy controls (n = 13, filled circles, P = .03, Figure 1A-B); however, it was normal in the group on treatment with TPO agents with increased platelet counts (open squares, P = .02, Figure 1B, and Table 1 and supplemental Table 1). In contrast to the CD19⁺CD24^hiCD38^hi B cells, the frequency of CD19⁺CD24^intCD38^int (mostly mature B cells) and CD19⁺CD24⁺CD38⁻ (primarily memory B cells) subsets in nonsplenectomized patients were not statistically different compared with healthy controls, and did not appear to be affected by platelet counts (supplemental Figure 2B-C).

Splenectomized patients, irrespective of platelet counts, had a higher frequency of CD19⁺CD24^hiCD38^hi cells compared with nonsplenectomized patients (Figure 1B, Table 1, and supplemental Table 1) and the frequencies of CD19⁺CD24^intCD38^int and CD19⁺CD24⁺CD38⁻ subsets in splenectomized patients were altered in comparison to those of healthy controls (supplemental Figure 2B-C). A marked deficiency of memory CD19⁺CD27⁺ B cells in some ITP patients was noted which as previously described was particularly pronounced in the splenectomized cohort^33,34  (Figure 1C, supplementary Figure 2).

Human Bregs mediate regulatory activity partly via IL-10–dependent mechanisms.^3,5  To determine whether B-cell IL-10 expression levels differ in ITP patients, we grouped the patients based on platelet counts of above or below 50 × 10⁹/L and compared the 2 groups to B-cell IL-10 expression levels in healthy controls. Some of the patients who were analyzed for phenotypic expression of B-regulatory markers were also assayed for B-cell intracellular IL-10 expression (see Table 2). In addition, patients on treatment with TPO agents who did not have a platelet response were included in the B-cell IL-10 expression analysis of the group with lower (< 50 × 10⁹/L) platelet counts, while patients in remission or off treatment with higher (> 50 × 10⁹/L) platelet counts were analyzed with the group that had elevated (> 50 × 10⁹/L) platelet numbers (see Table 2). As previously reported,⁵  we found rare IL-10–producing human B cells before stimulation (data not shown). Although cytoplasmic IL-10 expression was dramatically increased after stimulation with a B-cell activating agent, Toll-like receptor 9 (TLR9) agonist CpG oligonucleotides (CpG), the levels in total as well as CD19⁺ subsets were significantly less in ITP patients with low platelet counts than in healthy controls (Figure 2A-B, supplementary Figure 3, P < .03). IL-10 expression was equally low in splenectomized (n = 3) and nonsplenectomized patients (n = 4) with low platelet counts and did not appear to improve in patients with elevated platelet counts (n = 6, Figure 2B); secreted IL-10 levels in culture supernatants were not tested.

B-cell subpopulations activated through TLR and CD40 engagement can negatively regulate monocyte cytokine production during in vitro functional assays.⁵  Because of the low numbers of peripheral CD19⁺CD24^hiCD38^hi B cells that make their purification in sufficient numbers difficult, we analyzed the regulatory function of bulk B-cell populations in ITP patients (n = 7, see Table 3). Purified CD19⁺ cells activated using a previously described⁵  combination of CpG and sCD40L were cultured with monocytes that were stimulated with LPS to secrete TNF-α. Although TNF-α levels in stimulated monocytes from healthy controls (mean, 44% ± 2%) and ITP patients (mean, 53% ± 6%) when cultured alone were not significantly different (supplemental Figure 4, P = .07), suppression of monocyte TNF-α expression by activated B cells was less in patients with low platelet counts (mean, 56% ± 5%) compared with controls (mean, 21% ± 5%; P = .001; Figure 2C-D), suggesting that the ability of ITP B-cell compartment to control proinflammatory activity of other APCs is compromised. We also measured monocyte inhibitory activity of bulk B cells in a small number of patients (n = 4), one in remission and 3 on treatment whose platelet counts were above 50 × 10⁹/L and found an improvement in TNF-α inhibition (mean, 37% ± 5%; P = .04; Figure 2D), although it did not reach the levels found in healthy controls (P = .04).

In summary, we have observed a significant distortion of B-cell homeostasis in our cohort of patients with ITP which consisted mostly of patients on treatment with thrombopoietic agents but with various platelet count responses. Phenotypically, the frequency of peripheral CD19⁺ subsets based on CD24 and CD38 expression was significantly altered in ITP patients. We found impaired IL-10 response in stimulated B cells and a reduced B-cell suppressive activity with less dampening of monocyte activation in ITP patients with low platelet counts, consistent with a generalized immune-regulatory cell defect in ITP.^20–25  Importantly, an improvement, specifically in the frequency of CD19⁺CD24^hiCD38^hi subpopulation, previously described as Bregs,³  in nonsplenectomized ITP patients with increased platelet counts on treatment with TPO agents compared with those with low platelet counts, but off treatment, was observed. Although the observation that Bregs may improve in responders to TPO treatment needs to be confirmed in larger longitudinal studies, we were able to analyze 5 nonsplenectomized patients before and after therapy at the phenotypic level and found that indeed their CD19⁺CD24^hiCD38^hi frequency did improve as their platelet counts increased after receiving treatment (patients 1, 2, 4, 5, and 9, see Table 1). It should be noted that we cannot exclude an effect of drugs, namely the TPO agents, on the Breg compartment and studies that compare Bregs in TPO treatment responders versus nonresponders are needed to address this. In a small group of nonsplenectomized patients (n = 5) who were off treatment and had > 50 × 10⁹/L platelet counts, we also found a trend toward an increase in CD19⁺CD24^hiCD38^hi frequency compared with the cohort (n = 9) off treatment with low platelet counts (6.1% ± 0.8% to 9.4% ± 1.6%, P = .06). In renal transplantation patients, higher CD19⁺CD24^hiCD38^hi frequencies were associated with positive outcomes.⁶  One hypothesis is that increased frequency of CD19⁺CD24^hiCD38^hi subpopulation in ITP patients with higher platelet counts can lead to improvement in B cell–mediated immunoregulation, although direct proof of this will have to await the isolation and characterization of this subset in patients with low-versus-high platelet counts in patients off treatment. Nevertheless, we observed an increased B cell–mediated monocyte suppressive activity in 4 patients with platelet counts of > 50 × 10⁹/L compared with the group with lower counts (< 50 × 10⁹/L), indicating an improved B-regulatory compartment in ITP patients with elevated platelet numbers which we believe is independent of the effect of drug since there was roughly equal representation of patients on TPO agents in the group with low (< 50 × 10⁹/L) and high (> 50 × 10⁹/L) platelet counts. It should be stressed that future studies with larger and more focused patient groups are needed to confirm these findings and exclude the possible effect of the drug on the Breg compartment. Because CD40L is expressed on and in platelets,³⁵  and B-cell regulatory activity in vitro requires activation through CD40 engagement,³  one may speculate that platelets may directly contribute to improved Breg activity in vivo in ITP patients with increased platelet counts, although this may imply that up-regulation of CD40L, normally associated with activated platelets,³⁶  would stimulate Breg activity.

A deficiency of memory B cells has been previously reported in patients with ITP,¹³  and similar low numbers of peripheral memory B cells have been reported in patients with primary Sjogren syndrome,³⁷  as well as in patients with severe chronic sarcoidosis,³⁸  and in a subgroup of patients with common variable immunodeficiency with autoimmune manifestations.¹³  The underlying explanation for decreased memory B cells remains unknown, but it may be that because of continuous stimulation with platelet autoantigens, memory cells differentiate into plasma cells. Because naive B cells differentiate into either memory or plasma cells,³⁹  another possibility may be that there is more plasma cell differentiation and therefore less memory cells in ITP patients. The disturbances in the B-cell compartment were particularly pronounced in our splenectomized ITP cohort, possibly caused by impaired B-cell differentiation resulting from loss of germinal centers of secondary lymphoid organs.^33,34  Our splenectomized cohort were refractory ITP patients and had all been nonresponders to splenectomy. Most were being treated with TPO agents at the time of this study and yet regardless of platelet numbers, all had elevated frequency of the CD19⁺CD24^hiCD38^hi subset. It remains unknown whether the quality of transitional cells, including the CD19⁺CD24^hiCD38^hi subset, differ in splenectomized versus nonsplenectomized individuals and whether these cells are functionally active in our splenectomized patients. Interestingly, splenectomy responders have normalized T-cell receptor V-β repertoires, whereas nonresponders have abnormally expanded T-cell clones,⁴⁰  consistent with a state of immune dyregulation in nonresponders. In addition, some of our patient cohort, regardless of splenectomy status, had been treated several years before this study with the anti-CD20 monoclonal antibody, rituximab, which also depletes Bregs, and all but one had been nonresponsive. Based on studies with patients with SLE and rheumatoid arthritis, clinical remission with rituximab treatment is thought to be associated with preferential repopulation of Breg compartment relative to pathogenic B cells.⁴¹  We therefore speculate that long-term ITP responders to rituximab will have an early and persistent repopulation of Bregs and a delay in pathogenic B-cell reconstitution after rituximab therapy, whereas in nonresponders there will be a delay in reconstitution of Breg compartment. Interestingly, ITP rituximab responders were reported to have reduced T-cell clonality⁴²  and improvement in Treg numbers and function,²³  although their Breg compartment was not analyzed. It remains to be determined whether ITP rituximab nonresponders have particularly lower numerical or functional Bregs compared with responders or even those not previously treated with rituximab. Furthermore, future studies that analyze both the Treg as well as the Breg compartments of patients with ITP are needed to systematically address the relative importance of these regulatory cells in disease pathogenesis and response to treatment.

The potential clinical importance and utility of the findings reported here will have to await confirmation by completion of ongoing prospective, longitudinal studies. Combined with previous reports of defects in Treg compartment in ITP,^20–25  our data suggest that a generalized dysregulation in immune-regulatory networks may be present in ITP as in other autoimmune diseases. In mouse models, IL-10–secreting regulatory B cells promote differentiation of Tregs⁸  or their recruitment.⁹  Although the ability of human Bregs on Treg differentiation has not yet been demonstrated, it may be that altered Bregs in ITP patients are responsible for the compromised Treg compartment.^20–25  Decreased IL-10 secretion (Figure 2B) may affect Treg differentiation. Alternatively, because direct engagement of Bregs with Tregs was reported to be required for up-regulation of Foxp3 and CTLA-4, 2 molecules critical for Treg activity,⁴³  a reduced Breg compartment may directly impact Treg activity. Based on the current data we cannot distinguish whether the dysregulation is the trigger for autoimmunity or secondary to the ITP disease process. Longitudinal studies that monitor patients as the disease progresses are needed to address the in vivo relevance of Breg dysregulation in ITP. Nevertheless, our preliminary observation that as with Tregs,^23,32  the B-regulatory compartment has the potential to be normalized after therapy raises the possibility that analysis of regulatory B-cell subpopulations could serve as cellular biomarkers for specific defects in patients with ITP as with Treg-associated markers that might guide therapy in ITP patients in the future. This raises an important issue that manipulations that only increase Treg activity may be insufficient for treatments of ITP that hope to have lasting effects. It also brings up the question of how these 2 immunoregulatory cell types interact with each other and indicates that there may be a hierarchy among these regulatory compartments. Future studies are needed to analyze possible interactions of different immunoregulatory compartments and determine their relevance to disease and therapy in ITP.

This work was supported in part by National Heart, Lung, and Blood Institute grant HL096497-01 (K.Y.).

National Institutes of Health

Contribution: X.L. conceived the idea, performed research, analyzed data, and drafted the manuscript; H.Z. performed research, and analyzed and interpreted data; W.B. performed research, analyzed data, and edited the manuscript; N.B. and J.E. recruited patients and obtained their consent; M.A.H. compiled clinical data; J.B. selected and recruited patients and edited the manuscript; and K.Y. designed, directed, and wrote the manuscript.

Conflict-of-interest disclosure: J.B. receives clinical research support from the following companies: Amgen, Cangene, GlaxoSmithKline, Genzyme, IgG of America, Immunomedics, Ligand, Eisai Inc, Shionogi, and Sysmex. The family of J.B. owns stock in Amgen and GlaxoSmithKline. J.B. has participated in Advisory Boards for Amgen, GlaxoSmithKline, Ligand, Shionogi, and Eisai, and has also had a 1-day consult with Portola. The remaining authors declare no competing financial interests.

Correspondence: Karina Yazdanbakhsh, PhD, Laboratory of Complement Biology, New York Blood Center, 310 E 67th St, New York, NY 10065; e-mail: kyazdanbakhsh@nybloodcenter.org.