Introduction
Type 1 diabetes is an organ-specific autoimmune disease characterised by immune-mediated beta cell destruction in pancreatic islets, which results in deficient insulin production. Although T cells directly damage insulin-producing beta cells, B cells are key in this multifactorial process. Rituximab treatment (B cell depletion therapy) in individuals with type 1 diabetes delayed loss of C-peptide in the first year after diagnosis [
1]. Furthermore, B cell depletion in NOD mice restored normoglycaemia in a proportion of diabetic mice [
2,
3] and improved islet allograft survival rate [
4].
However, global B cell depletion may induce unwanted side effects [
5] and specific B cell targeting may be safer. In this context, rituximab specifically suppressed anti-insulin autoantibodies more than other autoantibodies [
6]. Animal studies have shown anti-insulin B cells are important in the pathogenesis of type 1 diabetes. A B cell transgenic mouse, expressing a fixed heavy-chain B cell receptor (BCR) with potential for insulin binding, had an increased frequency of anti-insulin B cells in a polyclonal repertoire (VH125) and accelerated diabetes onset [
7]. Conversely, the control heavy-chain transgenic mouse (VH281), with restricted insulin binding, had a reduced diabetes incidence [
7]. Furthermore, anti-insulin B cell depletion using the mAb123 antibody in NOD mice protected against spontaneous diabetes [
8]. Thus, autoreactive anti-insulin B cells play a particularly important role in the pathogenesis of type 1 diabetes.
Although self-reactive B cells undergo central tolerance in the bone marrow, autoantigen-specific B cells are found in the peripheral B cell repertoire, in a functionally silent or anergic state [
9]. Anti-insulin B cells (125Tg model) have impaired responses to both innate and adaptive B cell stimulators such as lipopolysaccharide and anti-CD40, respectively [
10]. However, these B cells can still present antigen and stimulate both naive and insulin-specific CD4 T cells [
11], suggesting that even in an impaired state, autoreactive B cells can still promote type 1 diabetes.
B1 B cells are important players in initiation of disease [
12] and are present early in the pancreas of NOD [
12,
13] and DO11xRIP-mOVA mice [
12‐
14], whereas established islet B cells have a more follicular phenotype [
14], although antigen specificity has not been investigated. Anti-insulin B cells have been clearly identified in pancreatic islets of NOD mice and are enriched at this site, compared with secondary lymphoid organs [
15]. The aim of our study was to track the fate of anti-insulin B cells in the pancreatic islets after anti-CD20 treatment by using a double transgenic NOD mouse. Here, we have focused specifically on the phenotype and functionality of insulin-binding cells in the pancreatic islets, a tissue that is not accessible in humans at specific times in the pathogenesis of diabetes.
Discussion
In this study, we describe novel anti-insulin B cell populations that reside in the pancreatic islets during type 1 diabetes development. Anti-insulin B cells are selectively recruited to pancreatic tissue during diabetes progression and upon entry these cells assume a unique CD138int phenotype. Furthermore, for the first time, we show that during global B cell depletion, anti-insulin B cells are not spared but are targeted in VH125.hCD20/NOD mice and, importantly, we have identified a key pattern of repopulation after resetting the B cell repertoire. This work highlights the need to further understand the dynamics of anti-insulin B cells.
Previous work performed in VH125Tg and V
H125
SD NOD mouse models supports the notion that anti-insulin B cells are functionally active, respond to mitogens and are effective APCs despite their tolerant state [
7,
11,
33]. Here, we show that anti-insulin B cells from our VH125.hCD20/NOD mouse model successfully present antigen to insulin-specific CD8 T cells. Consequently, anti-insulin B cells not only present to CD4 T cells [
11] but also successfully present to pathogenic insulin-specific CD8 T cells.
Insulin-reactive B cells escape immune tolerance in mice susceptible to type 1 diabetes [
8]. Indeed, we show that anti-insulin B cells are selectively recruited to pancreatic tissue in VH125.hCD20/NOD mice. Our data supports a previous study, using a different method of detection (insulin-specific mAb123-biotin), showing increased insulin-binding B cells in the pancreas of VH125Tg/NOD mice [
15]. While we have not addressed whether anti-insulin B cells are specifically recruited via a chemokine receptor, it is conceivable that anti-insulin B cells express high levels of C-X-C motif chemokine receptor 3 (CXCR3), a chemokine receptor involved in recruitment of lymphocytes in autoimmunity [
34,
35]. High levels of CXCR3 are expressed on other autoreactive B cell subsets [
36] and CXCR3 is involved in localisation of plasma cells [
37]. Further investigation is required to fully understand whether the enrichment observed is a result of recruitment or retention.
We reveal that anti-insulin B cells are unique, although heterogeneous, in pancreatic islets, with enrichment of CD138 and loss of their naive BCR isotypes (IgM/IgD). This suggests that some B cells enter the plasma-cell differentiation pathway. Although the expression of CD138 on pancreatic islet B cells has been described previously [
27], and more recently on insulin-binding B cells in VH125.NOD mice [
26], this is the first time the heterogeneity has been reported. Furthermore, CD138 intermediate expression coupled with a lack of Blimp-1 demonstrates that islet B cells are not terminally differentiated. This does not exclude the possibility that CD138
hi B cells have previously expressed an insulin-specific BCR but lost the BCR on terminal differentiation, [
38,
39] so are unable to detect and bind insulin. CD138
hiIgD
lo B cells remain MHC II positive. However, as MHC II is lost on terminally differentiated B cells, these islet CD138
hiIgD
lo B cells may also represent an early differentiated plasma-cell type [
37] or a short-lived plasma-cell [
40]. While loss of BCR expression would render B cells unresponsive to antigen, expression of MHC II would still allow antigen presentation in an inflammatory microenvironment, possibly driving beta cell destruction. Intermediate expression of CD138 is reminiscent of autoreactive anti-sm (ribonucleoprotein Smith) B cells described in the spleen of autoimmune mice [
41]. However, some CD138
int anti-insulin B cells that are still IgD competent may resemble IgM
lowIgD
high mature follicular B cells, a unique splenic follicular zone subset [
42]. It is clear CD138
int and CD138
hi are distinct populations, and their relationship and role in auto-inflammation is yet to be defined.
We note that wild-type NOD islet B cells resemble the B
ND (anergic naive) compartment in peripheral blood of humans, which are also IgM
lowIgD
+ in phenotype. This anergic B
ND compartment is enriched with a pool of autoreactive B cells [
43]. However, these B
ND cells are lost in individuals with newly diagnosed type 1 diabetes [
24] for reasons that are currently unclear. Our data support the notion that these B cells may have relocated to pancreatic islets and are thus lost from peripheral blood. Interestingly, co-expression of IgD and IgM promotes accumulation of anergic B cells and increased CD138 expression is associated with reduced IgM [
44]. Furthermore, a loss of IgD expression on B cells induces amplified CD138 expression [
44], possibly reflecting the response of anti-insulin B cells entering pancreatic tissue. In humans, CD138
+ B cells have been detected in islets that are positive for CD20
+ B cells in five out of 29 individuals studied [
45,
46]. This suggests either that there are similar migration patterns in human pancreatic islets or that B cells change in situ, the latter being consonant with our observations.
Our work in this diabetes model is consistent with earlier studies in arthritis [
32] showing successful depletion of peripheral autoreactive B cells using an hCD20 transgenic system. While a previous study suggested that all B cells downregulate CD20 upon islet entry [
27], we show this is not the case and that a heterogeneous population exists, which includes CD20
+ B cells. This disparity may be due to the different anti-CD20 mAb used. We detected some anti-insulin B cells that had lost CD20 expression and were spared from anti-CD20 treatment. However, this population is small and diabetes is delayed in a large proportion of mice and hence anti-CD20 treatment is clearly beneficial, at least temporarily.
The early insulin-positive B cell recruitment to pancreatic islets after anti-CD20 treatment is an important observation in the therapeutic use of B cell depletion. We demonstrate that insulin
+CD19
− B cells are increased in the pancreatic environment upon repopulation, indicating that during repopulation, pancreas-infiltrating anti-insulin B cells are recruited earlier and enter the plasma-cell differentiation pathway more rapidly. Alternatively, B cells (developing B cells [
47]) already expressing CD138 in the bone marrow exit prematurely after anti-CD20 treatment and track to pancreatic islets. Anti-CD20 treatment alters the islet microenvironment [
19] and may allow CD19
− anti-insulin B cells to proliferate in situ or receive increased survival signals. B cell activating factor (BAFF) improves the survival rate of autoreactive B cells [
48] and increased BAFF levels are seen in individuals after B cell depletion therapy [
49,
50]. Factors such as BAFF may influence the return of the autoreactive B cell repertoire and allow anti-insulin B cells to populate pancreatic tissue following anti-CD20 treatment.
The function of both CD138
int populations is of importance, particularly the enriched insulin
+CD19
− observed after anti-CD20 treatment. These cells may be functionally altered and may change the pancreatic environment. Whether early recruitment of anti-insulin B cells to the pancreas after global B cell depletion contributes to disease persistence or, alternatively, may in some way participate in the delay of diabetes needs to be addressed. This latter possibility is of particular interest, as recently CD138 (syndecan 1) has been identified as a hallmark of anergy [
44]. It should also be considered that the insulin
+ B cells we have defined have opposing roles. We acknowledge that while we would like to test these possibilities, functional assays on such small populations is highly technically challenging and would require many pancreas samples to be pooled from large numbers of experimental animals for each group, rendering this unfeasible with current technology. However, this would be an important investigation for the future. Nevertheless, our study highlights a unique phenotype of islet-infiltrating B cells, emphasising the need for greater understanding of autoreactive B cells and how they promote the development of type 1 diabetes, as well as providing novel insight to help with better design of more effective immunotherapies.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.