It is now well-established that regulatory B cells (Bregs) can dampen immune responses and play a role in maintaining immune tolerance. These immunosuppressive Bregs are generally named for the anti-inflammatory cytokines that they produce to exert their regulatory effects, and so a variety of Bregs have been identified. The cytokine most widely associated with Bregs is Interleukin-(IL-)10 (1) and thus has been the major focus of many studies into the failure of Bregs to suppress inflammation in autoimmune conditions. IL-10 independent mechanisms have been identified, including suppression mediated by contact of cell surface molecules (2, 3) or other soluble mediators such as the production of TGFβ (4) and IL-35 (5). However, currently there are no reports of alterations in these IL-10 independent regulatory B cell populations, either in number or function, in human type 1 diabetes; thus their contribution to type 1 diabetes remains an outstanding question.

In type 1 diabetes B cells are typically understood to play a pathogenic role in disease, likely through the production of inflammatory cytokines and presentation of autoantigens to T cells (6). This has been emphasized by the use of Rituximab in clinical trials and the observed temporary delay in the loss of C-peptide (7). However, studies of other autoimmune diseases have highlighted the essential role for regulatory B cells (8) and this has now been reflected in type 1 diabetes, although comparatively with fewer studies. Regulatory B cells in other autoimmune diseases, including diabetes, has recently been reviewed (9). It is imperative that we further understand the balance between effector and regulatory B cells in order to improve immunotherapeutic treatments targeting these lymphocytes, including utilizing Bregs as a therapeutic option. This review will focus on the emerging literature on Bregs and discuss their role in type 1 diabetes.

Studies in both human and mouse have contributed to identifying numerous IL-10-producing Breg subsets using a variety of immune markers, some of which overlap, to indicate a regulatory population. In humans, several Breg subsets enriched at different stages of B cell maturation, including immature B cells (CD24^hiCD38^hi) (10), memory B cells (CD24^hiCD27⁺ [B10]) (11) and plasmablasts (CD27^intCD38⁺) (12) have been identified. Similarly, in mice, various subsets have been identified in the transitional (13) and marginal zone (14) B cell compartments, including specific mouse subsets that parallel human B10 cells (11, 15) and human plasmablasts (12).

Other human regulatory B cell subsets have also been described including CD19⁺Tim-1⁺ B cells (16) and CD39⁺CD73⁺ Bregs (17), with equivalent subsets described in mice (18, 19). In addition, human CD25^hiCD71^hi B cells produce IgG4 and are designated as regulatory Br1 cells (20). However, these subsets have not yet been described in human type 1 diabetes. The diversity and identification of Breg phenotypes has been reviewed extensively (9, 21, 22). The range and variability in methods which induce IL-10-producing B cells, along with a lack of a key definitive marker, makes it difficult to define a Breg cell without assessing IL-10 production, as a key function. Therefore, the evaluation of IL-10-production during the differentiation and developmental stages of B cells is important, as demonstrated by Iwata et al. reporting the distinction between B10 cells and B10-progenitor cells (B10_PRO) (11). The different subsets of Bregs that have been assessed, specifically in studies of type 1 diabetes, is discussed (in Impaired Regulatory B Cell Mechanisms in type 1 diabetes) and Table 1 .

The heterogeneity of Bregs, both in phenotype and response to stimuli, and the absence of a definitive single marker (so far) has led to the hypothesis that any B cell can differentiate into a Breg depending on their prevailing environment, rather than a subset derived from a distinct lineage (21). Indeed, signals required for the induction or the promotion of regulatory B cells are the result of an activated inflammatory environment, including pro-inflammatory cytokines, engagement of Toll-like receptors (TLRs) and costimulatory signals (32, 33). This has been reviewed extensively (34). Certainly, evidence from mouse studies show that Bregs are induced in response to inflammation or autoimmunity (13, 35). Moreover, a number of cytokines are involved in promoting Breg responses, many of which have been associated with autoimmune disorders. In autoimmune diabetes a number of cytokines including IL-1β, IL-6 and Interferon (IFN)α, play a role in the development of disease and can contribute to pancreatic β cell death (36). The same cytokines, as well as IL-21, have been shown to activate or expand Breg function (33, 37). IFNα secreted from plasmacytoid DCs (pDCs), in combination with CD40 ligation, can induce IL-10-producing Bregs (37). B cells stimulated with cytosine-phosphate-guanine (CpG) dinucleotides in combination with IL-2, IL-6 and IFNα induced an enhanced IL-10 response (12). Furthermore, IL-1β and IL-6 can drive B cell IL-10-production and Breg differentiation (33). Interestingly, this raises the question of why then in some studies Bregs are numerically or functionally defective in autoimmunity that includes type 1 diabetes (see Impaired Regulatory B Cell Mechanisms in type 1 diabetes). One possible reason for this paradox could be explained by other mechanisms required for Breg induction, which are altered in autoimmunity ( Figure 1 ).

In a human study of SLE, the failed Breg expansion is attributed to elevated levels of IFNα produced from pDCs during disease, which drives plasmablast differentiation rather than Breg expansion (37). Therefore, it is suggested the concentration levels of cytokine are an important factor in Breg induction, and chronic exposure during inflammation can impair Breg frequency and function (37, 38). Type 1 diabetes, like SLE, is associated with an IFN signature. IFNα expression detected in the pancreatic islets (39) and IFN-associated genes are overexpressed in islets of individuals with type 1 diabetes (40). Additionally, an IFN transcriptional signature has been shown to be increased, even before the onset of human islet autoimmunity (41).

Both IL-21 and CD40 receptor engagement are required for the maturation and function of IL-10-producing B cells, a key study demonstrated in mice (42). Interestingly, naïve B cell responses to IL-21 are diminished in established human type 1 diabetes; however this response is enhanced in pre-diabetic individuals with multiple islet autoantibodies (43). Furthermore, CD4 T follicular helper (Tfh) cells in patients with type 1 diabetes have increased IL-21 production, compared to healthy donors (44, 45).

The importance of CD40: CD40L signaling has been noted in autoimmunity. For example, in autoimmune diabetes, the influence of CD40L blockade on the development of diabetes has been demonstrated in the NOD mouse model (46). This fundamental signaling pathway is important in both T and B cells. In people with type 1 diabetes, CD4^loCD40⁺ T cells (T_CD40) are expanded in peripheral blood (47). In another autoimmune disease, SLE, aberrant expression of CD40L in circulating B cells, in addition to T cells has been noted (48). Furthermore, reduced numbers of CD40⁺ B cells is observed in individuals with type 1 diabetes, compared to healthy donors; however, the levels of CD40 expression on B cells were not measured in this study (28).

Other mechanisms are necessary for the generation or expansion of Bregs and these include both adaptive and innate immune pathways. B cell receptor (BCR) signaling (32, 49), is diminished in B cells from individuals with established type 1 diabetes (25, 43). Signaling through TLR9 changes the frequency and function of IL-10 producing B cells in NOD mice; TLR9 deficiency specifically in B cells increased IL-10 producing cells and protected against diabetes (50). No direct study has demonstrated a mechanism that drives a Breg defect in type 1 diabetes in humans however, and this remains an outstanding question (see Discussion and Outstanding Questions).

Studies on the numerical and functional defects of Bregs have been described in various autoimmune diseases, including SLE, RA and MS and overall an inverse correlation between the frequency of Bregs and disease activity has been observed (10, 51). It should be noted, however, that studies have also reported either no differences or an increased frequency in these cells between autoimmune individuals and healthy donors (11, 25). Others have demonstrated different levels of CD24^hiCD38^hi Bregs in various autoimmune conditions, compared with healthy controls (52). This theme of contrary results is echoed in studies of type 1 diabetes, which are summarized in Table 1 .

It still remains unclear whether a defect or an impaired function of Bregs contributes to the development of diabetes or if the observed aberrant frequency and function is a result of chronic inflammation. Studies in the Experimental autoimmune encephalomyelitis (EAE) mouse model of MS has implicated Bregs in disease initiation rather than late-phase progression (53, 54). Moreover, in NOD mice, early treatment (5-6 weeks old) with BCR-activated B cells both delayed and reduced diabetes onset; however later treatment at 9 weeks of age only delayed onset of disease (55). Determining how Bregs contribute to the onset of type 1 diabetes will be of significance when considering immunotherapies targeted at B cells. Future real-time studies of regulatory B cells in islet autoantibody-positive individuals, who have not yet developed overt type 1 diabetes, would improve understanding of this.

B cells residing in pancreatic islets during inflammation contribute to the destruction of β cells, and consequently a loss in the secretion of insulin. Evidence for this direct pathogenic role has been shown by B cell depletion studies in the NOD mouse model, highlighting a reduction in effector T cell function inhibiting tissue-specific inflammation in treated mice (67, 68). In NOD mice, B-1a cells located in the pancreas, early in diabetes, play a role in initiation of disease (69). Furthermore, the observation of different profiles of insulitis in human pancreatic islets, with increased frequency of CD20 B cells correlate with a more progressive earlier diagnosis (58).

Previously, we have alluded to proposed interactions between regulatory B cells and the inflammatory pancreatic islet environment, and how Bregs can control inflammation (70). Islet-specific B cells in naturally-protected normoglycemic NOD mice have increased IL-10 and CD40 expression (28). More recently, we have corroborated this work and demonstrated B cells from naturally-protected NOD mice have an increased frequency of B cells expressing IL-10, CD80 and CD40 (61). In this study we also described an enrichment of CD19^intCD138^hiCD44^hiKi67⁺ dividing plasmablasts in naturally-protected NOD mice (61) a phenotype attributed to IL-10 production (12). Alongside this increase in regulatory B cells, a significant increase of CTLA4⁺FoxP3⁺ Tregs was also observed (61), possibly indicating some Breg-Treg crosstalk, which suppresses local pancreatic inflammation. However, it is unknown if this crosstalk is dependent on the expression of IL-10. It is possible that other IL-10-independent Treg induction by B cells may occur, as shown by the requirement for Breg expression of GITR ligand (71). Moreover, it is currently unclear if the altered pancreatic milieu in naturally-protected NOD mice is responsible for the induction of these regulatory immune cells, or a result of expanded IL-10⁺B cells in the periphery (60). IL-10⁺ B cells can be detected in the pancreatic islets of younger NOD mice after CD40 ligation along with a PMA/Ionomycin stimulation, albeit the frequency of IL-10⁺ B cells was very low (28, 61), and it is unknown if they have any role in controlling local β cell damage in vivo.

As discussed above, studies of regulatory B cells in type 1 diabetes are limited in comparison to other autoimmune diseases that include SLE, RA and MS, and thus lessons can be learned in order to extrapolate the findings to direct key research in type 1 diabetes. Finally, we discuss future and outstanding research questions that will advance the treatments of type 1 diabetes.

The complex picture described, so far, in diabetes and other autoimmune diseases may reflect the divergent role of Breg subsets in various disease settings. Different subsets of Bregs, based on their maturity, may be more influenced by the level of inflammation and disease stage of the individual. As previously shown, different immune profiles for B cell and T cell responses are dependent on disease stage or progression (43). It should also be noted that IL-10-producing B cells can also secrete TNF and IL-6 and so there is heterogeneity in Breg cytokine production (72).

Interrogating Breg: immune cell crosstalk will uncover aberrant regulatory feedback loops. Cell subsets like pDCs (37) or other DC subsets will reveal how Bregs dampen, or fail to dampen APCs. Other Breg studies highlight immunosuppressive mechanisms via invariant NKT cells dependent on the surface molecule CD1d (73). Other studies describe a feedback loop between T cells and B cells, via CD40:CD40L interactions, to develop regulatory function, which differentially regulate T cell proliferation and Th1 responses (74).

Determining if the defect in Breg frequency and/or function is a consequence of chronic inflammation or a contributor to the development of diabetes will have an impact on how B cell depletion therapy is exploited in individuals during various stages of disease progression. Furthermore, understanding if impaired Bregs contribute to disease due to the lack of immunosuppressive action or if Breg plasticity results in a further progression of disease under certain chronic conditions should be addressed.

So far, only a pan-B cell depletion approach has been trialed in type 1 diabetes (Rituximab) (7), and therefore we can only discuss preclinical studies that approach expanding Bregs in vivo or targeting a specific B cell population. Expansion of CD73⁺ regulatory B cells after treatment with a small molecule inhibitor that disrupts the Aicda-encoded activation-induced cytidine deaminase protein (AID) results in the inhibition of diabetes development in the NOD mouse (75). Conversely, AID deficiency in the NOD mouse model can accelerate type 1 diabetes development (76) and therefore the role of AID in diabetes progression requires further investigation. An additional B cell-targeted therapeutic approach is to selectively deplete effector B cells preserving regulatory B cells; however this is complicated by the lack of a definitive Breg marker. Interestingly, targeting of B cells via the blockade of the B cell activating factor (BAFF) induced an increase of IL-10⁺ B cells and diabetes protection (77). Furthermore, in this study, anti-CD20 treatment depleted this IL-10-producing B cell population, suggesting that Bregs are more sensitive to deletion during anti-CD20 treatment (77). This Breg sensitivity may have contributed to the limited success of the Rituximab clinical trial (7). However, as discussed above, a deeper understanding of Bregs during the development of type 1 diabetes is needed to harness and develop successful B cell targeted immunotherapies.

Overall, the pathogenesis of type 1 diabetes is complex and multi-stage, and requires a number of pathogenic cell types that give rise to the development of disease. Equally, it is clear that balanced against these pathogenic cells are regulatory cells, that include both T and B cell subsets. Defining the roles of these less-understood Breg subsets will provide important information to be further studied in humans with the aim of increasing therapeutic opportunities.

JB wrote and edited the manuscript. FSW edited the manuscript. All authors contributed to the article and approved the submitted version.

JB is supported by an Independent Fellowship funded by Research England’s Expanding Excellence in England (E3) fund via EXCEED. Medical Research Council (UK) grant MR/K021141/1 was awarded to FSW.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.