Invariant natural killer T cells balance B cell immunity

Abstract

Invariant natural killer T (iNKT) cells mediate rapid immune responses which bridge the gap between innate and adaptive responses to pathogens while also providing key regulation to maintain immune homeostasis. Both types of important iNKT immune responses are mediated through interactions with innate and adaptive B cells. As such, iNKT cells sit at the decision-making fulcrum between regulating inflammatory or autoreactive B cells and supporting protective or regulatory B cell populations. iNKT cells interpret the signals in their environment to set the tone for subsequent adaptive responses, with outcomes ranging from getting licensed to maintain homeostasis as an iNKT regulatory cell (iNKT_reg) or being activated to become an iNKT follicular helper (iNKT_FH) cell supporting pathogen-specific effector B cells. Here we review iNKT and B cell cooperation across the spectrum of immune outcomes, including during allergy and autoimmune disease, tumor surveillance and immunotherapy, or pathogen defense and vaccine responses. Because of their key role as influencers, iNKT cells provide a valuable target for therapeutic interventions. Understanding the nature of the interactions between iNKT and B cells will enable the development of clinical interventions to strategically target regulatory iNKT and B cell populations or inflammatory ones, depending on the circumstance.

1 |. INVARIANT NATURAL KILLER T CELLS

Invariant natural killer T (iNKT) cells are a unique population of T cells capable of rapid, early cytokine responses upon activation. Invariant NKT cells are restricted to the non-polymorphic, MHC class I-like antigen-presenting molecule, CD1d. CD1d mediates selection for development of iNKT cells in the thymus and presents glycolipid antigens from pathogens for iNKT cell activation in the periphery. iNKT cells express a semi-invariant T cell receptor, with a conserved Vα chain and a limited repertoire of Vβ chains.¹ iNKT cells are uniquely selected by CD1d expressed by double-positive thymocytes, to progress through three stages of iNKT development, stage 0/1 (CD44^negNK1.1^neg), stage 2 (CD44^pos, NK1.1^neg), and stage 3 (CD44^pos, NK1.1^pos).² Much like conventional T cells, once they reach the periphery and inhabit secondary lymphoid organs and other tissues, they are comprised of a diverse population of functional subsets including iNKT1, iNKT2, iNKT17, iNKT_FH, and iNKT10 defined by their transcription factor expression and cytokine production.^3–6 iNKT1 cells are found predominantly in the spleen and CXCL10-rich environment provided by the liver; can be defined by their expression of the CXCL10-ligand, CXCR3, and the transcription factor T-bet; and predominantly produce the cytokine IFNγ when activated. iNKT2 cells favor production of IL-13, IL-5, and IL-4 which contributes to CD8⁺ T cell differentiation in the thymus, express predominantly GATA3, and in addition to the thymus, can be found in the lung where they contribute to allergic airway responses. iNKT17 cells express the transcription factor RORγt, chemokine receptor CCR6, and can be found in increased frequencies in the skin, lung, and peripheral lymph nodes where they produce IL-17, IL-21, and IL-22.⁷

While cytokine production and transcription factor expression define these subsets, the specific signals which dictate the differentiation of iNKT cell subsets remain to be defined. Differences in TcR signal strength following engagement of antigenic ligands suggest signal strength may be a key component of this differentiation. For example, a surrogate of TcR signal strength following TcR engagement, Nur77 expression, is greater in iNKT2 than iNKT17, which is greater still than iNKT1 cells.^8,9 TcR signaling also induces expression of PLZF, the lineage-specific transcription factor expressed by most iNKT cell subsets.^10,11 Consistent with the differences in TcR signal strength implied by the NuR77 expression levels, PLZF is also differentially expressed by iNKT subsets, with iNKT2 expressing the most, iNKT 17 an intermediate amount, and iNKT1 cells expressing the least.^12,13 An alternative theory proposes differential cytokine-mediated signals drive commitment to iNKT subsets, as they do for conventional T cells, with IL-15 driving iNKT1 populations, IL-7 driving iNKT17 cells, and IL-25 responsible for iNKT2 and iNKT17 cells.⁴ However, differential cytokine receptor expression by the iNKT subsets must be dictated by an earlier signal, so these two theories may not be mutually exclusive.

Additional important iNKT cell subsets have been characterized specifically in peripheral organs: iNKTreg, iNKT_FH, and iNKTFreg in germinal centers of the spleen and both NK1.1^pos and NK1.1^neg subsets of PLZF^neg iNKT cells in adipose tissue. In the adipose tissue, NK1.1^neg cells produce predominantly IL-10 and are positive for the transcription factor E4BP4, similar to the iNKT10 population characterized after robust iNKT activation ¹⁴; however, the adipose iNKT cells do not express PLZF.^15–17 Adoptive transfer studies with splenic iNKT cells which home to adipose tissue in the recipient and convert to PLZF^neg iNKT cells suggest development of the adipose iNKT subset depends exclusively on the milieu encountered in that tissue.¹⁵ However, mutations in the TcR sequence which alter CD1d binding also result in selective accumulation of adipose-resident iNKT cells,¹⁸ suggesting TcR signal strength and consequent PLZF reduction via alternative mechanisms can also influence the development of the adipose iNKT population. Both iNKT10 and adipose iNKT cells may also be the product of chronic activation, as evidenced by increased activation markers CD69 and Nur77 as well as exhaustion marker PD-1.^14,17 Interestingly, adipose iNKT cells have recently been characterized to be comprised of two PLZF^neg populations, one NK1.1^pos and another NK1.1^neg, which do not interconvert.¹⁵ iNKT regulatory cells (iNKTreg) are another iNKT subset not typically found in naive animals, but which have been found to develop during chronic inflammation. Similar to their T regulatory cell counterparts, iNKTreg cells express Foxp3 and critically depend on TGFβ for their development.¹⁹ They were initially described following glycolipid activation, and can regulate autoimmune disease through CD95 and GITR, but still express iNKT-specific traits including the transcription factor PLZF.¹⁹ iNKT follicular helper (iNKT_FH) cells are CXCR5⁺ bcl6⁺ iNKT cells which provide similar help to B cells as their conventional T_FH counterparts. iNKT_FH cells have been localized to the B cell follicle in the spleen where they produce cytokines IL-21, IL-4, IFNγ, and others to help enhance B cell humoral responses including antigen-specific antibody production, class switching, and affinity maturation.^20,21 Other related iNKT subsets have subsequently been identified in the spleen, including iNKTFreg which express markers of both iNKT_FH cells (CXCR5, bcl6) and Foxp3.²²

2 |. B CELLS

Another lymphocyte population capable of early, innate responses which is counter-regulated and influenced by iNKT cell subsets is B cells. The early innate lymphocyte immune response is not only important for the immediate response to invading pathogens, but also to initiate and shape the following adaptive immune response. Recognition of pathogen-associated molecules by pattern recognition receptors such as Toll-like receptors (TLR), NOD-like receptors (NLR), or scavenger receptors (SR) on antigen-presenting cells (APCs) is followed by internalization of the pathogen and antigen presentation on MHC molecules.²³ This in turn activates T cells, and the subsequent T cell–dependent activation (TD) of B cells leads to antibody production as a result of the germinal center (GC) reaction.

The T cell–dependent response is typically against protein antigens and requires a signal from the B cell receptor (BCR), called signal one, where B cells get activated and pick up antigen from follicular dendritic cells (FDCs). This is followed by antigen processing and migration to the border between the B and T cell area in secondary lymphoid organs. Here the B cell receives signal two by presenting antigen to T cells that have been previously activated by APCs in the T cell zone. This interaction leads to reciprocal activation and initiates class switch as well as differentiation of B cells into non-affinity matured memory cells and extrafollicular plasma cells.^24,25 The initial proliferation of T cells, now differentiated into T_FH cells, and B cells, also starts the GC reaction to promote further affinity maturation. Affinity maturation is dictated by the level of access to T cell help available for B cells that mutated their BCR through upregulation of activation induced deaminase (AID).²⁶ In the GC, the B cells which acquire higher affinity through mutations also acquire that ability to take up more antigen and in effect also get more T cell help. In the spleen, plasma cells are also generated at the early stage of activation before a fully developed GC emerges and these B cells migrate to the bridging channels in close proximity to the T cell zone and red pulp. Here, the TD B cells produce an early antibody response that is important to combat infection.²⁷ In addition, unswitched memory B cells are also generated at this early stage of activation. Upon reactivation, these unswitched memory B cells can be recruited for somatic hypermutation during re-exposure of the antigen.²⁸ Or alternatively, they can quickly become plasma cells to generate a more potent quick boost of antibodies during an infection.

As a complement to the TD response, some B cell responses to non-protein antigens can be rapid and T cell–independent (TI). T cell–independent responses were initially detected when B cells were activated by certain antigens in athymic mice.²⁹ The TI antigens can be further divided according to whether they require signaling from the BCR to elicit a response.^30,31 This division was initially characterized in CBA/N mice which have a defect in the signaling kinase Btk, first discovered in humans with X-linked immunodeficiency.^32,33 Antigens like LPS or CPG are called TI antigens of type I (TI-I) and do not require signaling through the BCR. Instead, these antigens provide signal 1 via TLR engagement. Alternatively, TI antigens of type II (TI-II) are large polysaccharide molecules such as Ficoll and Dextran with repeated antigenic determinants and the ability to activate complement. For these antigens, signal one is strong enough to bypass the need for further recruitment of activation signal two from T cells through B cell antigen presentation. Unlike TD responses, TI-II responses typically produce a strong extrafollicular response and are characterized by a relatively higher presence of polyreactive antibodies.

TI responses can also be supported by myeloid cells that provide additional activation signals (reviewed in 34). These myeloid-derived costimulatory boosts include production of IL-21, the TNF family members BAFF and APRIL, and cognate interaction through CD40. The TI-II category of help includes the activation of neutrophils by IL-10, which drives a B-helper neutrophil (N_BH) phenotype.³⁵ N_BH cells support B cell activation and promote survival and differentiation including upregulation of AID.³⁵ In a similar manner, DCs can support TI responses by production of APRIL and it was shown that monocyte-derived DCs can support production of IgA in mucosa-associated lymphoid tissues (MALTs).³⁶ Similarly, human macrophages produce BAFF after activation in vitro.³⁷ This shows that the TI response is also in close connection to the TD response and will influence the balance of the B cell response and direct the B cells into the different paths for GC selection, memory or plasma cell formation. See Figure 1 for a review of different types of B cell responses and the relative contribution of helper cells.

FIGURE 1.

Categorizing B cell responses as they relate to the influence of help from T cells

The TI B cell response is primarily mediated by the specific B cell subtypes B1a/b and marginal zone B cells (MZB). Compared to the more numerous follicular B cells (FOB), MZBs and B1 subsets display a limited BCR repertoire capable of cross-reactivity between different antigens that can be both foreign and self. Examples of these particular antigenic determinants include oxidation-specific epitopes found on oxidized low-density lipoproteins (oxLDL) and dying apoptotic cells.^38,39 MZBs in both humans and mice appear after birth and they develop both in germ-free, MHC, or T cell–deficient mice, suggesting that they do not need T cell help or foreign antigen to develop.⁴⁰ However, they fail to develop in mice that are deficient for CD19 suggesting that signals through the BCR are necessary. It has been shown that BCR signaling upregulates Notch2 which provides a key positive signal required for selection into the MZB pool.^41,42 MZB cells also fail to develop in mice lacking molecules needed for cytoskeletal changes connected to migration.^43,44 This is possibly to their function in transporting antigen to the follicle. This type of migration may also be needed to inhabit the niche of MZBs and, in effect, acquire the signals needed for their distinct phenotype.^45,46 The MZB and B1 B cell subsets are responsible for the production of natural serum antibodies which are enriched for polyreactivity, which is important in the early part of an immune response but can also be a source for potentially pathogenic autoantibodies.^47,48 The innate-like B cell subtypes display both phenotypic and migratory differences and inhabit specific micro-anatomical niches. Of these, the MZBs that reside in the marginal zone have been shown to interact with N_BH cells and also have the highest CD1d expression of all naive subpopulations. B1 B cells also express CD1d and primarily inhabit pleural and peritoneal cavities and maintain small populations in secondary lymphoid organs.^49–51 After activation, B-1 B cells can express CD1d and some degree of expression has been detected on other naive populations as well as on short-lived plasma cells and cytokine producing B cells including IL-10-producing regulatory B cells (Breg).^52,53 In humans, these innate B cell subtypes also exist even though the phenotype and development of B1 cells has been greatly debated.^49,54 On the other hand, MZBs are present in humans, and although they are more numerous and recirculate, they are not confined to the spleen-like murine MZBs. Human spleens have a different structure and do not have a distinct marginal zone–like mice; thus, recirculation of B cells is different.⁵⁵ However, human MZBs are dependent on NOTCH signaling in a similar manner as in mice.⁵⁶

3 |. GLYCOLIPID PRESENTATION BY B CELLS

Expression of CD1 molecules is regulated by multiple transcription factors and regulatory elements. In humans, the SP1 transcription factor regulates expression, whereas in mice, members of the ETS family of transcription factors regulate the expression of CD1d specifically in B cells.^57,58 Also, the murine regulatory region contains retinoic acid response elements and it has been shown that stimulation with retinoic acid increased CD1d expression in B cells.^52,59 Antigen presentation by B cells is closely linked to uptake of antigen via the BCR for presentation on MHC class II to conventional T cells. Subsequently, T_FH cells recognizing antigen in the context of MHC II give the second signal for B cells to go on with differentiation and assure that the B cells and T cells of the GC are specific for the same antigen even though the epitopes are different. CD1d is loaded for presentation in endosomal and lysosomal compartments through which it traffics from the cell surface and back again.⁶⁰ In these compartments, CD1d can exchange its glycolipid ligands either with endogenous or exogenously derived lipids before returning to the surface.⁶⁰ Thus, in B cells foreign lipids can be loaded on CD1d following BCR-mediated uptake. This has also been shown to take place and has been used to generate antigens for NKT cell activation where a B cell epitope mediating uptake via the BCR has been fused to a glycolipid to be presented on CD1d.^61–63 Using these BCR-targeted antigens for presentation on CD1d, it was shown that presentation was enhanced several 100-fold compared to glycolipids alone. Also, the dominating antibody subclasses produced following immunization with glycolipid antigen were IgM, IgG3, and IgG2c and MZBs were more effective for targeted glycolipid presentation compared to FOB. As an alternative to antigen-specific uptake, glycolipids can also be taken up for presentation via the LDL receptor that binds to lipid and ApoE-containing complexes in DCs and B cells in a BCR-independent manner.⁶⁴ This links glycolipid presentation to the metabolic needs and fitness of the cell and directly to lipid metabolism.

To this day, most studies on CD1d-mediated presentation have been done on DCs or macrophages. In these two professional APCs, it has been shown that CD1d traffics to lysosomes guided by the AP-3 adapter complex.⁶⁵ As a consequence, AP-3-deficient mice have higher CD1d on the surface on DCs since knocking this protein out prevents internalization. However, this is not true for B cells suggesting that CD1d loading and recycling is regulated differently in these cells. Since B cells are specifically equipped to present antigen taken up via the BCR and also endosomal pattern recognition is linked to this it is likely that adapter systems are involved. In connection to pattern recognition, members of the scavenger receptor family defined by the capacity to bind oxidized glycolipids have been shown to mediate presentation of glycolipids by CD1d.⁶⁶ Of these, CD36 and SR-B1 are expressed by B cells, with CD36 especially highly expressed by MZBs. As MZBs are in close contact with the circulation in the spleen, a CD36-mediated pathway is likely to play a role for lipid presentation, especially by the MZB subpopulation of B cells. CD1d expression drops drastically once B cells take on a GC phenotype, suggesting that at this stage, they are less prone to be regulated by NKT cells or require a stronger ligand presented on CD1d to be activated or selected.⁶⁷

4 |. INNATE B CELL RESPONSES AND INKT CELLS

Glycolipid antigens have been shown to elicit B cell responses independently of help from conventional T_FH cells. These responses are a result of cognate (innate) interactions between NKT cells through different pathways.⁶⁸ For this innate arm of the response, CD103⁺ DCs presenting the glycolipid agonist αGalactosylceramide (αGalCer) support B cell responses via DC expressed CD40 expression and IL-12-mediated activation of NKT cells to become NKT_FH cells.^20,69 NKT_FH cells share many of the same differentiation traits with T_FH cells in that they both upregulate CXCR5, PD-1, CD28, ICOS, and Bcl-6 and form cognate interactions with B cells in a cellular synapse that requires SAP signaling.^20,69 Initial in vivo studies found iNKT cell help for B cells required CD40L, B71/2, and IFNγ (but not IL-4) suggesting the help provided by iNKT cells was similar to that provided by conventional T_FH cells.⁶³ Both helper-supported responses are also dependent on IL-21 production.²¹ There are other strong similarities between iNKT_FH and T_FH cells. For instance, after antigen activation, iNKT_FH cells have been identified in B cell follicles ⁷⁰ and specifically in early ongoing germinal centers.⁷¹ As with conventional T_FH cells,⁷² iNKT_FH cells may also support retention of higher affinity GC B cells by producing BAFF during ongoing GC reactions.^22,73–75

Immunization with the haptenated glycolipid NP-αGalCer induces an innate, cognate interaction to occur where only iNKT_FH cells are expanded, not T_FH cells. NP-αGalCer induces iNKT_FH cells to provide costimulation and induce a rapid extrafollicular antigen-specific B cell response with production of IgG and IgM.⁵⁵ However, affinity maturation is not as potent when B cells are activated by NKT_FH cells and they generate limited memory and long-lived plasma cells.⁷⁵ This suggests that even though the response is TD, it is innate by nature.⁷⁵ While DCs are required for initial activation of iNKT cells, there is a cooperative requirement for iNKT:B cell cognate interaction in order to maintain iNKT_FH cells and iNKT_FH-dependent GCs. In B cell–deficient mice, immunization with αGalCer does not result in iNKT_FH cells, suggesting iNKT maintenance depends in part on interaction with B cells.²⁰ Similarly, adoptive transfer of SW_HEL CD1d^−/− B cells to WT recipients immunized with HEL-αGalCer demonstrated that antigen-specific GCs did not develop, even in the presence of normal iNKT_FH cells, when the antigen-specific B cells are not capable of antigen-specific uptake, αGalCer presentation by CD1d, and cognate help from iNKT cells.²⁰ Galli, Dellabona, and Abrignani used antigen plus αGalCer immunized MHC II^−/− mice or BM chimeras to demonstrate the critical requirement for antigen-specific B cells to express CD1d and CD40 in order to form antigen-specific GCs, produce increased antigen-specific class-switched antibodies, and expand iNKT_FH cells when immunized with a glycolipid antigen.^69,76 Other studies found a role for ICOS in iNKT: MZB cell interactions.⁷⁷ And as noted earlier, NKT cells also require SAP expression to properly mediate cognate help for B cells, much like conventional T_FH cells.⁶⁹ Thus, cognate interactions between B:iNKT cells are required for GC B cell development and iNKT_FH cell maintenance, but not for initiating iNKT_FH cell expansion. The high CD1d expression by MZBs is matched with a lower threshold for activation suggesting that cognate interactions are more likely to be the dominant means for NKT cells to regulate innate MZB cell responses. However, as the expression level of CD1d is lower on FOBs and absent on GC B cells, this non-cognate pathway is more likely to be at play when adaptive FOB and GCB cells are recruited to an antibody response.

Despite many similarities, iNKT_FH help for B cells also differs in some key features as compared to conventional T_FH help. Interestingly, B cell help provided exclusively by iNKT_FH cells does not produce humoral memory in the same way that help provided by conventional CD4⁺ T cells does.^20,21,75 First, iNKT cell help for B cells drives different antibody isotype profiles than help provided by conventional T_FH cells. B cells activated by NP-αGalCer and helped exclusively by iNKT cells produced primarily antigen-specific IgG3 and IgM, while conventional T-D antigen NP-KLH/αGalCer or protein/αGalCer conjugated particles helped by both iNKT and T_FH cells produced predominantly IgG1 and IgG2.^61,63 Furthermore, antigens engaging primarily iNKT cells as a source of help (OVA-αGalCer or NP-αGalCer) induced GCB cell development earlier than GCB cells induced in the presence of T_FH cells.^20–22 Exclusive iNKT_FH cell help expands GCB cells to be detectable by day 3 but they diminished by day 6. In fact, the GCB cell response induced by cognate iNKT cell help is reminiscent of T-independent B cell responses⁷⁸ where T-independent antigens drive early, large GCs which prematurely involute in the absence of T cell help. Physiological germinal center maturation likely depends on T cell signals which prevent centrocytes from undergoing apoptosis.⁷⁸ As yet unidentified T cell signals rescue germinal center B cells from apoptosis and induce a proportion of germinal center B cells to differentiate into centroblasts or memory B cells. In the absence of appropriate T cell signals, germinal center B cells fail to become centroblasts and will not produce memory B cells or PCs. Thus, premature involution of germinal centers and lack of memory B cells or PCs after cognate iNKT cell help are consistent with a lack of T-cell signals to sustain germinal centers. Wermeling et al⁶⁷ demonstrated that antigen-activated B cells significantly down-modulate CD1d expression upon entry into the germinal center. We noted that the same B cells upregulate MHC II during residence in the GC, which may leave them posed to receive T_FH signals, but unable to receive iNKT_FH signals, and thereby unable to sustain an iNKT-driven true adaptive GC B cell response [Yates, Leadbetter unpublished]. In contrast, non-cognate antigen engages both iNKT_FH and T_FH cells, and induces a later, modestly sustained germinal center B cell expansion that is consistent with germinal center B cell expansion patterns common to protein-specific T_FH cell help.²²

5 |. ADAPTIVE B CELL RESPONSES AND INKT CELLS

Non-cognate interactions are characterized as iNKT cells activated by glycolipid plus protein to promote activation of DCs and generation of iNKT_FH cells as well as potent T_FH cells.³⁵ As expected, only non-cognate immunization with αGalCer plus protein antigen is able to induce expansion of peptide-specific T_FH cells, and T_FH cells are critical for iNKT_FH-helped long-term humoral memory. For example, when CD4⁺ T cell–deficient MHC class II^−/− mice are immunized with peptide plus αGalCer, they form germinal centers and initiate a primary antibody response but do not maintain long-lasting secondary antibody.⁷⁵ By comparison, cognate iNKT_FH help for B cells fails to induce much affinity maturation or long-term B cell memory.^20,21 Thus, iNKT_FH cells may initiate GCB cell responses, but CD4⁺ T_FH cells are the cells primarily responsible for sustaining non-cognate humoral memory. However, iNKT cells may contribute to alternative IgM or TI B cell memory if antigens are delivered with a high valency or as components of a particle, but details of this mechanism remain to be defined.⁷⁹ With these features, the innate form of help provided by cognate iNKT cell engagement with B cells can be placed in the context of other categories of TI and TD B cell help. iNKT_FH cells provide similar cognate help to B cells, but the outcome is different. iNKT_FH help drives an early, aborted GC B cell response, class switch with modest affinity maturation, and no memory B cell response. As such, the innate help provided by iNKT cells can be characterized as T-dependent type II or TD-2.^34,79

6 |. INKT CELLS AND B CELLS IN AUTOIMMUNE DISEASE

B cells can have both protective and pathogenic roles in systemic autoimmune diseases. As an example, depletion of B cells after disease onset using antibodies reduced the disease in a model of atherosclerosis.⁸⁰ On the other hand, removal of the spleen aggravated the disease and transfer of B cells but not T cells to the mice reversed this showing that B cells conferred protection. We have recently found an explanation for this in connection with natural antibodies. In addition to the pathological response from FOB cells, we also discovered that atherosclerosis induced a protective B cell response originating in the marginal zone.⁸¹ This was driven by production of natural antibodies that cleared oxLDL from the system and driven by activation of the inflammasome.

Consistent with the dual role of B cells in atherosclerosis, a similar picture has emerged in human autoimmune diseases, including SLE, where autoantibodies drive the disease by generating pathological immune complexes. On the other hand, regulatory B cells, defined by their production of IL-10, were shown to ameliorate autoimmune disease.^82,83 The same is true for iNKT cells, as it was found that modeling the serum IL-18 elevation noted in several autoimmune diseases by injecting systemic IL-18 induced a self-reactive antibody response negatively regulated by iNKT cells.⁸⁴ However, if IL-18 was combined with stimulating glycolipid αGalCer, the regulatory function of iNKT cells was lost and instead they took on a iNKT_FH phenotype and supported the autoimmune response.⁸⁵

A role for iNKT cells in the K/BxN disease model was suggested by the observation that arthritis was attenuated in CD1d KO and Jα281 KO mice, both deficient in NKT cells.⁸⁶ The mechanism was thought to involve iNKT-mediated suppression of transforming growth factor β1 (TGF-β1) production in the joint tissue, which in turn was dependent on release of IL-4 and interferon γ (IFNγ) from the iNKT cells.⁸⁷ Furthermore, it was shown that CD1d KO mice exhibited less severe arthritis and that arthritis could only be restored with transfer of iNKT cells from WT mice and not with transfer of cells from FcγR KO mice. These findings indicate that binding of IgG to FcγRIII on iNKT cells in the joint induces their activation and participation in the induction of arthritis.

Natural killer T cells can also drive Breg responses. For example, αGalCer can induce an early expansion of MZBs that includes production of IL-10, which shows that they have a preference to support innate and early B cell activation capable of suppressing autoimmunity.²² Also, in a feedback like manner, Bregs can induce iNKT cells to produce IFNγ which modifies the balance of cytokine production by iNKT cells toward lower Th1 and Th17 type responses, which in effect decreases experimental arthritis.²² In this model, mice lacking CD1d on B cells had more severe disease, consistent with a contribution for cognate presentation by this type of activated B cell to the iNKT regulatory response. Furthermore, SLE patients display lower numbers of NKT cells in circulation and the level of CD1d expression on B cells is decreased.⁸⁸ Mechanistically, this has been linked to increased internalization of CD1d and suggests that the negative regulation of NKT cells is lost in SLE patients. This is likely disease-driven as treatment with rituximab depletes B cells and restores both NKT cells and B cell CD1d levels. Correspondingly, NKT cell numbers are also lower in patients with active multiple sclerosis, whereas they return when patients are in remission.^89,90 These data are consistent with the loss of a protective role for iNKT cells and recruitment of autoreactive B cells to the GC to generate more pathogenic antibodies in their absence.

Key features that decide whether an antibody response will be pathogenic or protective are the affinity of the antibodies and the subclass of the Fc-portion of the antibodies. If the antibody switches to a subclass that is more prone to drive pathology, disease can be exacerbated.^91,92 Thus, the innate boost of natural antibodies and specificities induced by iNKT cells is typically protective and this, in combination with induction of Breg cells, can prevent autoimmunity. Furthermore, iNKT expansion of Breg cells suppresses the switch and recruitment of B cells producing pathogenic antibodies as has been shown both for anti-dsDNA antibodies as well as for rheumatoid factors in SLE and RA.⁹³ This is also true in animal models as we have shown that injection of apoptotic cells to break B cell tolerance allow for a B cell response that can be protective.^67,94 However, in the absence of iNKT cells, breaking B cell tolerance instead becomes pathogenic, similar to SLE patients.⁶⁷

B cells and iNKT cells also play a role in the systemic autoimmune disease type 1 diabetes (T1D) which is a disease characterized by destruction of the insulin producing β-cells of the pancreas. Although the disease pathology is primarily driven by conventional T cells infiltrating the pancreas, it has been shown that B cells specific for islet antigens can be detected before disease onset.⁹⁵ Also, depletion of B cells using anti-CD20 antibodies prevents and even reverses disease development in susceptible mice.^96,97 Still, even though autoantibodies are a good biomarker in T1D in humans, in vivo studies suggest that they are not drivers of disease. In mice where B cells are deficient in the ability to secrete antibodies, there was no difference in disease development. This suggests that other functions of B cells, most likely cytokine production and antigen presentation to pathogenic T cells, are important for the supportive role for B cells in disease development.⁹⁸ In contrast, a protective role for Breg cells has been shown in T1D and production of IL-10 from these cells can suppress pathogenic CD8 T cell responses.⁹⁹ This is in line with findings in humans where T1D patients have been shown to have reduced proportion of Breg cells in circulation.¹⁰⁰ Connecting these B cell responses to the role of iNKT cells, it has been shown that injection of αGalCer can protect susceptible mouse strains (NOD) from T1D.^101–103 In this model, presentation of αGalCer by B cells seems to be an important contributing factor, suggesting that this could be one of the mechanisms conferring protection.¹⁰⁴ The protective role of iNKT cells is also evident in NOD mice which have defects in numbers and function of iNKT cells.¹⁰¹ Confirming the protective role, transfer experiments of iNKT cells protect recipients from T1D and furthermore, crossing NOD mice with TCR transgenics increasing iNKT cell numbers confers protection.¹⁰⁵

In summary, the polarization of iNKT cells and their influence on B cell responses can have different outcomes depending on the type of B cell response they are promoting. On one hand, if the iNKT cells support B cells in their function to produce autoantibodies or drive pathogenic T cell responses, iNKT cells will support disease development. On the other hand, iNKT cells can regulate beneficial boosts of natural antibodies that are capable of reducing disease-driving auto-antigens and activate Breg cells to form a suppressive alliance with B cells to protect from autoimmune disease. However, currently, B cell depletion using anti-CD20 antibodies (rituximab) is being used for treatment purposes for several autoimmune diseases including MS and SLE.^106,107 This shows that removing the dominating function of B cells to get reactivated and produce autoantibodies is enough during later stages of disease when treatment is applied. It is likely that any beneficial regulation by iNKT cells will have been lost at this point because tolerance has been broken and the disease is being perpetuated by pathological memory responses.

7 |. INKT AND B CELLS IN RESPONSE TO INFECTION

Invariant natural killer T cells play diverse roles in maintaining homeostasis in many peripheral organs, but they can also be activated to make key contributions to immune defense. B cells are not typically tissue resident, so while systemic iNKT surveillance for infected B cells includes peripheral tissues and tertiary lymphoid structures, iNKT support for pathogen-specific B cells seems to be largely restricted to secondary lymphoid organs. This may correlate with differences in helper vs lytic functions of specific subsets of iNKT cells present in spleen, LN, or Peyer’s Patches, the nature of the cytokine milieu, or the relative type or mechanism of antigen presentation. iNKT cells may control viral infections of B cells through recognition and direct cell lysis of infected B cells. In fact, human iNKT cells are reduced in number and/or function during multiple infection-induced B cell lymphoproliferative disorders, including EBV infection,^108,109 EBV-triggered disease subsequent to XIAP or XLP deficiency,¹¹⁰ and Castleman disease as a consequence of HHV-8 infection.¹¹¹ The cognate nature of iNKT control of B cells during these disorders is supported by in vitro studies demonstrating that EBV-infected B cells can specifically activate iNKT cell IFNγ production (even in the absence of exogenous antigen).¹⁰⁸ However, B cells rapidly lost CD1d expression during ongoing EBV infection and depletion of iNKT cells from culture led to increases in EBV-infected B cells and viral titers,¹⁰⁸ suggesting that CD1d downregulation is an effective iNKT immune evasion strategy employed by EBV during B cell infection.

Alternatively, B cells may be helped to produce pathogen-specific antibody as a means to control infection. In rare cases, iNKT cells are activated through their cognate recognition of specific glycolipid antigens derived from a pathogen, including Borellia Burgdorferi, Streptococcus pneumoniae, Sphingomonas sp, and possibly Helicobacter pylori.^112–115 In contrast, there is one example of this type of recognition by iNKT cells mediating an enhanced pathogenic B cell response in mice. In the case of primary biliary cirrhosis induced by infection with N aromaticivorans, the pathogen contains a B cell antigen, pyruvate dehydrogenases complex E2, and an iNKT cell antigen, a-galacturonosylceramide. Infection with N aromaticivorans leads to iNKT-dependent production of high-affinity, class-switched antibodies recognizing PDC-E2 which initiate organ damage similar to human primary biliary cirrhosis.¹¹⁶

More frequently, iNKT cells are activated by CD1d, IL-12, and IL-18 from pathogen-activated DCs¹¹⁷ and provide non-cognate, cytokine-mediated help to other immune cells including B cells. In one example, iNKT cells are critical for early bacterial clearance in the lung during the pneumoniae model of Streptococcus pneumoniae. iNKT cells are thought to contribute to bacterial clearance via IL-13-mediated help for high-affinity IgM produced by B1a cells.¹¹⁸ iNKT cells also mediate early bacterial clearance and subsequent protection against S pneumoniae through IFNγ production, which recruits neutrophils via TNF and MIP2.^119,120 B cells can also rely on early IL-4 secreted from interfollicular iNKT cells during viral infection (Influenza, Vaccinia, Zika) to seed germinal centers,¹²¹ so iNKT-deficient mice form poor GCs and make reduced antibody responses to viral infection. Similar reductions in MHC II–dependent and MHC II–independent antigens were seen with CD1d-deficient mice during infection with Plasmodium berghei.¹²² Early iNKT IL-4 secretion does not depend on cognate interactions with B cells, but rather early IL-18 and CD1d-mediated interactions with LN resident CD169⁺ macrophages.¹²¹ An additional example of an iNKT cell– mediated virus-specific B cell response comes from infection with HSV-1.¹²³ In this example, iNKT-deficient mice produce less IgM and IgG during HSV-1 infection, specifically the IFNγ-induced IgG2c and IgG3 subclasses, consistent with a loss of serum IFNγ. In a related model system, IgG1 switching by B cells in the Peyer’s Patches of iNKT transnuclear mice was similarly found to be dependent upon iNKT IL-4, not cognate interactions via CD1d.¹²⁴ The frequency of iNKT cells in the Peyer’s Patches of these mice varied depending on the mouse facility, implicating a role for commensal gut bacteria in expansion or maintenance of this population and subsequent IL-4 production. In summary, any TLR or PAMP/DAMP-engaging pathogen can elicit an iNKT cell response via non-cognate CD1d/cytokine activation. As such, broad, early, innate activation of iNKT cells could be important for many aspects of protective immune responses against almost all pathogens and does not limit the iNKT-mediated immune protection to only glycolipid antigen–containing pathogens.

Vaccines are now specifically being developed which harness non-cognate iNKT cell help for vaccines against human pathogens using glycolipid activation as a successful strategy in murine models, although no glycolipid-adjuvanted vaccines have yet reached humans in the clinic. In vitro studies with human iNKT cells suggest that while both CD4⁺ and CD4⁻CD8⁻ iNKT cells help B cells to proliferate, it is the CD4⁺ subset that provides better help for humoral antibody production.⁷⁶ One approach to engaging iNKT cells for help against pathogen infection is to combine a glycolipid agonist with current vaccine components to enrich the protective response. In some cases, activation of iNKT cells boosts only CD4 or CD8⁺ T cell responses,^125–128 but in other cases co-immunization with αGalCer and antigen can lead to increases in pathogen-specific antibodies and increased protection. For example, combining an iNKT agonist glycolipid with an HIV-DNA vaccine encoding env and gag proteins increases CD4 and CD8 T cell responses, as well as specific antibody titers.¹²⁹ Similar increases in virus-specific antibody response and protection from disease are observed when iNKT-targeting glycolipid adjuvants are co-injected with influenza proteins.^130–132 Co-encapsulation of αGalCer and target antigens recruits iNKT cell help to enhance class switching and humoral immune responses to T-independent polysaccharide antigens, such as those found in Streptococcus pneumoniae.⁷⁹ Although it remains to be seen if αGalCer enhanced antibody responses improve protection against S pneumoniae infection, the class-switched recall response induced by αGalCer-polysaccharide liposomes suggests it is likely.⁷⁹

8 |. INKT AND B CELL CONTRIBUTIONS TO INFLAMMATORY DISEASES

As is evident during infection, iNKT cells can contribute to general antibody production during chronic inflammation. Their capability to rapidly produce copious amounts of IL-4 in response to inflammatory cytokines induced by TLR ligands, positions them as early initiators of innate B cell responses. Studies from Yoshimoto and colleagues used adoptive transfer studies of iNKT cells to show that early, rapid IL-4 production elicited from iNKT cells induces IgE class switch and production.^133,134 CD1d KO mice reveal that iNKT cells are not exclusively required for this early IgE production, other Th2 cells also contribute,¹³⁵ but the iNKT-derived IL-4 can enhance antibody responses, either by increasing the numbers of iNKT cells using Vα14 transgenic mice, or activating the iNKT cells with repeated administration of IL-18 to mimic chronic inflammation.^133,136 Relatedly, Umetsu’s group found that glycolipid activation of iNKT cells in MHC II ko mice lacking conventional CD4⁺ T cells induced IL-4 and IL-13 expression and increased IgE production,.¹³⁷ While the role of iNKT cells in asthma induction or regulation remains controversial,¹³⁸ the increase in IgE is consistent with previous studies demonstrating a role for iNKT cells’ contribution to basal IL-4 levels.

In addition to supporting IgE production through basal IL-4 production, we have more recently demonstrated that iNKT cells are also capable of negatively regulating B cells, usually autoreactive B cells. Specifically, we found that iNKT cells normally regulate a splenic response that activates MZBs to produce IgE, IgM and IgG in response to chronic IL-18 administration in vivo.⁹⁴ The MZB population was greatly increased in iNKT-deficient mice,⁹⁴ and further study revealed that iNKT cells are licensed to take on this regulatory role by CD1d engagement with neutrophils and then go on to restrict expansion of these B cells via the perforin and CD95/CD178 pathways.⁸⁴ Interestingly, activating iNKT cells with a robust glycolipid antigen during IL-18-mediated chronic inflammation can over-ride this regulatory phenotype and drive them to an iNKT_FH phenotype which promotes autoimmune B cell responses instead.⁸⁵ In addition, this regulatory role of iNKT cells could be at work to keep a balanced IgE level at baseline as we found that in CD1d-deficient mice, IgE levels in serum increase over time without any immunization.⁹⁴ Reduced iNKT cell numbers are also found in humans in the primary immunodeficiency hyper IgE syndrome, which is due to a defect in cytokine signaling through STAT3, although a direct connection between the two has not been described.¹³⁹ Thus, lack of iNKT cells could be linked to auto-IgE in inflammatory diseases with high inflammatory cytokines, including IL-18, that reprogram iNKT cells. But also, in a steady state situation, perhaps iNKT deficiency can be connected to allergic diseases. In atopic eczema (AE), we found that IL-18 is increased in serum and correlates with disease severity, IgE level, and a skewed phenotype and reduced numbers of iNKT cells.¹⁴⁰ A similar phenotype could be induced in vitro when incubating human NKT cells with IL-18. NKT cells producing IFNγ and IL-4 are also part of the cellular infiltrate in the skin lesions of AE together with B cells.¹⁴¹ It has been speculated that allergens can also be presented on CD1d and in turn through cognate interactions with B cells regulate antigen-specific IgE responses [reviewed in 142]. For DCs, it was shown that glycolipids derived from pollen extracts could be presented to activate iNKT cells in vitro.^143,144 Also in a model for ragweed allergy, mice deficient in iNKT cells had reduced specific IgE and inflammatory response in the lung.¹⁴⁵

The regulation of IgE responses by iNKT cells is interesting and should be investigated more both in connection to allergy but also autoimmunity. Unfortunately, studies in some of the original mouse strains engineered to lack CD1d were hampered by the fact that an additional unintended mutation led to a coincident depletion of MZBs in these mice.^94,146 Furthermore, in the course of deleting the Jα18 segment of the TCR to create iNKT-deficient mice also deleted part of the upstream Jalpha locus resulting in a hampered conventional TCR repertoire.¹⁴⁷ It is likely that, with more recently created strains of mice, some of the early studies in these disease models should be redone with the hope of resolving some of the discrepancies in these studies relating to iNKT cell regulation of B cells. The more recent strains of mice will also be helpful to investigate the true role for Type II NKT cells as the perturbed TCR repertoire will not cloud the results.^148,149

While we now appreciate the important lead played by iNKT cells in the thymus, spleen, LN, and other primary lymphoid organs, much remains to be determined about their contributions to maintaining homeostasis and controlling or contributing to inflammation in other peripheral lymphoid tissues. For example, iNKT cells are enriched in both adipose tissue and liver, but we are just beginning to understand the mechanism of iNKT effect in those tissues. iNKT cells play a critical role in maintaining homeostasis in lean adipose tissue,¹⁵⁰ in part through their unique adipose phenotype marked by expression of the transcription factor E4BP4 and IL-10 production.¹⁵¹ Adipose iNKT cells are also unique in their unexpected absence of expression of PLZF, a transcription factor common to most innate-like lineage cells, including iNKT cells.¹⁵¹ It is now appreciated that the NKT population in the adipose is comprised of two distinct subsets, NK1.1^pos and NK1.1^neg, which maintain homeostasis utilizing two different mechanisms. The NK1.1^pos restricts macrophage populations through IFNγ production, while the NK1.1^neg population secretes IL-10 to maintain other regulatory populations like CD4⁺ Treg cells.¹⁵ It remains to be seen if iNKT-derived IL-10 contributes to maintaining regulatory B cells in lean adipose, but this certainly seems likely.

Interestingly, fat-associated lymphoid clusters (FALCs), a site for B cell proliferation and germinal center development, have been described to develop in murine visceral adipose tissue during chronic inflammation.¹⁵² FALCs are initiated by the action of TNF-expressing myeloid cells on adipose stromal populations, but their formation is dependent on CD1d-restricted iNKT cells. Thus, iNKT-supported FALCs serve as a site for antigen-specific B cell retention, expansion, and germinal center maturation, which can be induced by peritoneal immunization with antigen.¹⁵² Thus, given the array of interactions described for the iNKT cells and the various B cell populations in many primary immune tissues, their interactions in adipose tissue are also likely to take on the dichotomous regulatory dynamic during the homeostatic lean state as well as the chronically inflamed obese state. Many questions remain relating to the nature and mechanism of iNKT and B cell interactions in lean and obese adipose tissue may extend to the related peripheral tissues such as liver, pancreas, and omentum.

9 |. INKT AND B CELLS IN CANCER

Invariant natural killer T cells are a powerful mediator of natural tumor surveillance, with iNKT-deficient mice being more susceptible to tumor induction.^153,154 The tumor rejection capabilities of iNKT cells can be direct, with iNKT cells lysing tumor cells that are presenting altered-self glycolipids in CD1d, or iNKT cells can indirectly enhance the tumor rejection capabilities of other leukocytes.¹⁵⁵ iNKT-mediated tumor rejection can also be independent of CD8⁺ cytotoxic T cells.¹⁵³ INKT cells produce inflammatory cytokines which can activate NK cells and can be enhanced to inhibit tumor angiogenesis,^156,157 but they also contribute to regulation of myeloid-derived tumor-suppressor cells.¹⁵⁸ In one example, iNKT cells also provide cognate help for B cells which recognize a tumor-specific glycolipid antigen, N-glycolyl-GM3, to increase human tumor antigen-specific antibodies.¹⁵⁹ Certain human B cell leukemias and non-Hodgkin’s lymphoma also express reduced levels of CD1d, which may contribute to reductions in protective immunosurveillance normally mediated by iNKT cells.^160–162 Enhancement of CD1d expression on CLL B cells by retinoic acid treatment restores iNKT cell lysis in vitro.¹⁶³ Multiple myeloma cells also downregulate CD1d, and iNKT cells are depleted from circulation during the course of disease.^164,165

In addition to homeostatic tumor surveillance of B cell lymphomas, iNKT cells can also be directly harnessed for tumor control by glycolipid activation delivered via a number of different possible formulations. Initial studies administered soluble αGalCer to activate iNKT cells as a means to harness their tumor rejection capabilities, but these studies revealed that overactivation of iNKT cells results in iNKT anergy and DC lysis.^166,167 Furthermore, glycolipid presentation by B cells was initially thought to be more likely to induce iNKT anergy than antigen presentation by DCs, perhaps because of less efficient or different costimulation.^168,169 Subsequent studies tested administration of αGalCer-loaded DCs and found that this approach circumvents the problem of anergy and activates iNKT cells for modest tumor control, even in humans.^170–173 Specifically, αGalCer presented by DCs can engage iNKT cells to provide licensing signals to improve the DC antigen presentation capabilities for both glycolipids and proteins in vivo and in vitro, increasing production of IL-12 and IL-18 cytokines and upregulating costimulatory molecules such as CD40, OX40, and CD70,^174–177 thereby enhancing the DC-mediated activation of other related immune cells including NK cells, CD8⁺ T cells, CD4⁺ T cells, and B cells [reviewed in 178]. B cell antibodies do not naturally play a dominant role in tumor rejection, but B cells can be non-specifically expanded during the course of immune activating immunotherapy. For example, following adoptive transfer of αGalCer-loaded DCs and activation of human Vα24⁺ VB11⁺ NKT cells, B cells, in addition to NK and T cells, increase in frequency, likely via increases in serum IFNγ and IL-12.¹⁷²

The tumor stroma and the cells that reside here have come into focus in the recent years with the success of T cell–targeted therapies. Also, many studies have been launched to try to improve this novel treatment and overcome the overall anti-inflammatory environment that the tumor needs to grow and escape treatment. Studies have focused especially on myeloid cells such as tumor-associated macrophages and myeloid-derived suppressor cells. B cells also infiltrate the tumor and can produce cytokines including IL-10 from regulatory subsets.⁵³ Also, other cytokines can be produced including Th1 type cytokines like IFNγ, TNF-α, and IL-12 from Be-1-activated B cells. This is contrasted to Be-2 polarization of B cells that gives more IL-2, IL-13, and IL-6 production.¹⁷⁹ Of these, we have shown that Breg cells have high CD1d expression and can be induced and regulated by iNKT cells. Since iNKT cells with the right phenotype to do this are present, it could be a way that NKT cells suppress immunity in the tumor. Whether Be-1 cells or Be-2 B cells can be induced or regulated by iNKT cells in the context of tumor stroma remains to be investigated. It should be stated also that the cytokine production by B cells is not fixed and multiple signals will influence their cytokine production including those produced by iNKT cells.

10 |. FUTURE PERSPECTIVE

These studies place iNKT cells at the fulcrum between regulating inflammatory or autoreactive B cells and supporting protective or regulatory B cell populations (summarized in Figure 2). iNKT cells must consolidate and interpret external signals from the cytokine milieu, additive signals from other inflammatory or regulatory populations, and antigen presentation signals to determine the appropriate effector function for the situation. Given their early innate response time, iNKT cells set the tone for subsequent adaptive responses, with outcomes ranging from getting licensed to maintain homeostasis or being activated to serve as supportive helpers of pathogen-specific effector B cells. iNKT cells then use cytokine and chemokine signals to direct the overall response by most other leukocytes during naive homeostasis or inflammatory responses. One key mechanism that needs study in this respect is how and when B cells present glycolipids on CD1d since the interaction with NKT cells can both result in blocking and enhancement of the B cell response. Just as the conductor chooses the tempo and tone for each symphony played by the orchestra, the iNKT cell sets the response for the immune players, urging on a protective humoral response or reigning in an out of control autoreactive or pathogenic one. Thus, the nature of the iNKT interaction with B cells is shaped by many factors, including the iNKT subset, central or peripheral tissue residency, cognate CD1d-interaction vs non-cognate activation, and sterile or pathogen-induced inflammation. As such, iNKT cells remain an influential and intriguing target for therapeutic interventions. However, one has to be careful given the dichotomous nature of iNKT cells, since restoring iNKT cells does not necessarily restore B cell homeostasis unless one can specifically target regulatory iNKT populations, rather than inflammatory ones. With our current understanding of iNKT regulation of B cells, the interactions between these two important cells which dictate protective vs pathogenic polarization state, characterized by differential cytokine production and effector function, merit further study.

FIGURE 2.

iNKT cells mediate divergent outcomes for B cell immune responses