Introduction
Systemic lupus erythematosus (SLE), the prototype of a systemic autoimmune disease, is characterized by the production of pathogenic autoantibodies that directly or indirectly contribute to the pathogenesis of SLE, resulting in cell destruction and inflammation [
1,
2]. NZB/W mice spontaneously develop a disease closely resembling human SLE. We have shown before that these mice develop both long-lived and short-lived autoreactive plasma cells, and that long-lived plasma cells (LLPCs) contribute significantly to the production of pathogenic autoantibodies [
3]. These LLPCs are able to induce nephritis when transferred into immunodeficient mice [
4]. As they are refractory to immunosuppressive drugs (for example, cyclophosphamide, dexamethasone and a combination of the two) and B cell depletion, they represent a therapeutic challenge in the treatment of SLE [
3,
5,
6].
Autoantibodies are detectable years before the clinical onset of SLE in humans [
7], and by the age of only 4 weeks in NZB/W mice ([
3,
8] and unpublished data). Some of these autoantibodies are produced by LLPCs since they do not disappear upon treatment of humans or mice with drugs like cyclophosphamide [
3,
9] or rituximab [
10,
11]. However, it remains controversial when this population of refractory LLPCs is established in the course of the disease. We have previously shown that a population of autoreactive LLPCs exists in the spleen and bone marrow by week 24 of life [
3]. Whether such population is established early in disease pathogenesis and no longer formed later, when constant generation of short-lived plasma cells (SLPCs) may become a hallmark of pathology [
12], remain unclear. Alternatively, it has been proposed that a constant generation and turnover of the LLPC pool may be sustained by B cell hyperreactivity [
13,
14], but also this hypothesis remains to be elucidated. This is valuable information in order not to miss an ‘LLPC window of opportunity’ at the beginning of the disease. Moreover, although interesting studies showed that B cells are able to repopulate the plasma cell-deficient bone marrow [
15], it remains rather unclear whether in autoimmunity LLPCs may be replenished from autoreactive memory B cells after therapeutic depletion of these cells.
Here, we show that LLPC generation starts very early in NZB/W F1 mice, long before clinical onset of disease. Then, LLPC counts in the spleen plateau after about 10 weeks, but those in the bone marrow and inflamed kidney increase over lifetime. When PCs are eliminated by bortezomib [
16], LLPC counts recover within 15 days in both the spleen and bone marrow. Thus, for persistent elimination of autoreactive LLPCs, existing LLPCs must be depleted (for example, by a cycle of bortezomib), and their regeneration must be prevented by maintenance therapy. Maintenance therapy could be directed at eliminating precursor cells or preventing their activation. Here, we used a combination of bortezomib with cyclophosphamide as a model to demonstrate that, in contrast to bortezomib or cyclophosphamide alone, this combination therapy achieves sustained elimination of LLPCs.
Discussion
Here, we showed for the first time that the generation of autoreactive LLPCs in SLE-prone NZB/W mice starts very early in life, long before the onset of disease. Already mice at the age of 4 weeks had anti-dsDNA-secreting plasma cells resistant to cyclophosphamide in both bone marrow and spleen. In the bone marrow, autoreactive LLPC counts increased at a stable rate over life without reaching a plateau, even in six-month-old mice. In the spleen, LLPC numbers increased only in the first 12 weeks and then remained stable over time. This demonstrates that autoreactive LLPCs are constantly generated in NZB/W mice and that, later in life, they accumulate in the bone marrow and inflamed kidney but not in the spleen. These findings pose a therapeutic challenge since LLPCs are resistant to conventional immunosuppression (for example, high-dose cyclophosphamide and/or dexamethasone) or B cell-depletion strategies [
3,
5,
6]. They can be efficiently ablated by proteasome inhibition with bortezomib [
16]. However, after depletion of LLPCs with bortezomib, the LLPC counts recovered within 15 days reaching the levels of untreated mice. Thus, for maintained ablation of autoreactive plasma cells, the regeneration of LLPCs must be blocked as well. Here, we showed that depletion of plasma cells by bortezomib in combination with a maintenance therapy to prevent the regeneration of autoreactive LLPCs results in persistent ablation of autoreactive LLPC in NZB/W mice.
Are these findings of any relevance for human autoimmune diseases?
SLE autoantibodies are considered to be pathogenic in human SLE [
1], and some of them can be produced by LLPCs especially in refractory patients, as determined based on their resistance to cyclophosphamide and B cell-depleting therapy as with rituximab [
6,
9]. Moreover, their titers increase over time in active disease [
1], indicating continued generation of autoreactive LLPCs. Remarkably, plasma cell generation has been identified as a marker of active disease [
21-
23]. Furthermore, elimination of all plasma cells including LLPC and B cells (that is, their precursors) by anti-thymocyte globulin (ATG) followed by autologous stem cells transplantation leads to long-term remission in SLE patients [
9]. All of this evidence strongly suggests that the continuous generation of autoreactive LLPCs, their important role in disease pathogenesis and the need of targeting B cells and plasma cells for the therapeutic elimination of the autoreactive LLPCs, are modeled by NZB/W mice.
Notably, the dynamics of generation and maintenance of autoreactive LLPCs is not only determined by the rate of generation, but also by the capacity of the body to support LLPCs in the long run. It has been shown that the number of plasma cells in human and mouse bone marrow is determined by the number of chemokine (C-X-C motif) ligand 12 (CXCL12)-expressing stromal cells, which organize survival niches for individual LLPCs [
24]. The frequency of such stromal cells in the bone marrow is approximately 1%; accordingly, the physiological frequency of bone marrow plasma cells is also about 1% [
25]. Thus, about 10
9 and 10
6 LLPCs can be hosted in the bone marrow of healthy humans or healthy mice, respectively [
12,
25]. Here we found that the maximum capacity to support LLPC survival in the spleen is reached after only 12 weeks, in line with our previous findings [
3]. In the bone marrow, this number was reached after 29 weeks (the end of the observation period). Importantly, LLPCs accumulate in the inflamed kidney of SLE mice at a later stage of the disease, confirming that nephritic kidneys can provide survival niches for LLPCs [
14,
18,
20], increasing the capacity to support LLPC survival during disease. In humans, the situation may be slightly different in that a patient’s capacity to host LLPCs in the bone marrow may be reached before disease onset or early in disease. Nevertheless, the new LLPCs can be generated and maintained efficiently, likely by newly formed plasmablasts outcompeting LLPC for their survival niches [
12,
26] or, most probably, by homing of new autoreactive LLPCs in new niches in inflamed tissues as shown by us and others for SLE mice [
14,
18,
20].
Our findings have relevant clinical implications. As discussed above, the existence of autoreactive LLPCs is a therapeutic challenge. Considering the very early accumulation of autoreactive LLPCs in the bone marrow and spleen of NZB/W mice, our data suggest that a ‘clinically relevant window of opportunity’ for preventing the accumulation of autoreactive LLPCs would exist only for the kidney. However, persistent activation and accumulation of autoreactive LLPCs in the bone marrow may cause relapses in patient with SLE even after long periods of clinical inactivity [
6,
27]. Therefore other therapies aimed at targeting LLPCs are needed. LLPCs can be eliminated efficiently by ATG [
9], anti-lymphocyte function-associated antigen 1 (LFA1) plus anti-very late antigen-4 (VLA4) [
15], transmembrane activator and calcium modulator and cyclophilin ligand interactor-immunoglobulin (TACI-Ig) [
28] and bortezomib [
16] also in advanced stage of the disease. However, therapeutic ablation via these approaches has two big disadvantages. First, it is not selective for autoreactive LLPCs, but also eliminates protective LLPCs in all cases [
9]. Second, in lasting immune reactions and in autoimmune reactions, LLPC counts quickly recover within four weeks [
15] or two weeks, as shown here for NZB/W mice and suggested recently for SLE patients [
29]. Moreover, continued elimination of plasma cells cannot be a preferred therapeutic option since that would imply long-term immunodeficiency with the complete absence of humoral immunity and increased infection-related mortality [
30]. Here, we suggest an alternative approach combining plasma cell ablation therapy with follow-up treatment to suppress the regeneration of autoreactive LLPCs. In the case of bortezomib, this also would have the additional benefit of reducing unwanted side effects like neurotoxicity and thrombocytopenia [
31]. It could be argued that also this combination therapy with agents targeting plasma cells and B cells may promote an indiscriminate ablation of auotoreactive as well as protective antibodies with the obvious caveats regarding a higher risk of infection. Notably, we described that bortezomib treatment of patients with SLE induced a greater reduction in pathogenic antibody titers (anti-dsDNA antibodies 58.7% reduction) than protective ones (for example anti-tetanus antibodies 29.2% reduction) ([
29] and Alexander
et al.). Moreover, the administration of cyclophosphamide for immunosuppression, as used here, is only a proof of principle. Other more fitting options are available for patients (that is, depleting B cells with anti-CD20 or targeting B cell differentiation into plasma cells and survival with anti-BAFF). In particular, combined LLPCs targeting and anti-BAFF therapy (that is, using the approved drug belimumab) might be a first efficient way to eliminate LLPCs and on the other hand to interfere with their regeneration and persistence, with the advantage to contrast the increased level of BAFF described after B cell depletion [
32,
33]. Notably, the use of belimumab instead of a complete B cell-depletion therapy (for example, using rituximab) could help to preserve protective memory B lymphocytes promoting the regeneration only of the protective LLPC compartment. Indeed, treatment with belimumab is associated with significant reductions in the numbers of transitional, naive and activated B cells, as well as CD20 + CD138+ plasma cell precursors (plasmablasts) [
34-
36]. Conversely, the number of memory B cells and T cell is preserved after belimumab therapy indicating that these cells are independent of BAFF for survival [
37]. Consistent with the preservation of memory B cells and T cells, belimumab treatment does not substantially affect preexisting anti-pneumococcal or anti-tetanus toxoid antibody levels with similar rates of serious and/or severe infections as compared with the placebo-treated group [
34,
38-
40]. These results, together with the described greater reduction of autoreactive antibodies after bortezomib in SLE, let us speculate that belimumab combined with bortezomib treatment might not compromise the immune response to infection dramatically and indiscriminately. Finally, the combination of LLPC ablation with immunosuppression would also open options for the selective recovery of protective humoral immunity, for example, by vaccination or transfer of autologous, protective plasma cell precursors. Therefore, our results strongly suggest that when transferring such plasma cell depletion strategies to humans, combining plasma cell ablation with an efficient, preferably selective, ablation of the precursors of autoreactive LLPCs could be useful.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
AT carried out the depletion experiments, drafted and wrote the manuscript and conceived the studies. LK provided the kidney data, helped analyze the data and critically revised the manuscript. IMM and CV performed the BrdU pulse experiments, analyzed the data and drafted that part of the manuscript. QC provided kidney data and ELISPOT data and critically revised the manuscript. TA participated in the design of the study, critically discussed the data and helped to revise the manuscript. KVM participated in the design of the study, helped in analyzing the data and critically revised the manuscript. RAM participated in the study design, helped to draft the manuscript and critically revised the manuscript. AR and FH helped to design the study, drafted and wrote the manuscript. BFH designed the study, participated in the data generation (BrdU pulse chase, ELISPOT), analyzed the data, drafted and wrote the manuscript. All authors but IMM read and approved the final manuscript. IMM could not approve the final version of the manuscript as unfortunately he passed away before the completion of the manuscript.