Background
ICOS is an inducible costimulatory receptor expressed on activated T cells that is a member of the CD28-B7 family of costimulatory molecules [
1‐
4]. ICOS binds to the ligand B7h [
5] (also known as LICOS [
6], ICOSL [
7], GL50 [
8], B7RP-1 [
9], and B7-H2 [
10]), expressed constitutively on the cell surface of resting B cells and dendritic cells (DCs) [
5,
9,
11,
12], both of which can regulate T cell-dependent antibody responses. Studies of B7h
−/−and ICOS
−/− mice have demonstrated the requirement of the ICOS-B7h receptor-ligand pair in germinal center formation, class switched antibody production and antibody affinity maturation [
7,
13‐
15], but the potentially distinct roles of ICOS costimulation mediated by B7h-expressing B cells and DCs in the regulation of antibody responses have not been well defined.
ICOS signaling can promote IL-4 production, leading to Th2 polarization of differentiating CD4+ T cells [
16,
17], but can also enhance production of a variety of cytokines in other Th subsets that have already differentiated, including augmentation of IFN-γ production in Th1 cells [
18‐
20]. Thus, ICOS signaling in T cells can have different effects on immune responses depending upon the cellular context of ICOS-B7h interactions. In T cell-dependent antibody responses, ICOS-expressing activated CD4+ T cells can make contact with B7h-expressing antigen presenting cells in several distinct contexts, with the potential for regulation of different aspects of the response through ICOS signaling during each interaction. DCs can contact recently-activated CD4+ ICOS+ T cells in T cell zones [
21], and DCs in germinal centers can interact with ICOS+ Tfh cells [
22]. Recently activated B cells interact with cognate activated CD4+ T cells at the border of the T cell and B cell zones in lymph nodes and spleen prior to germinal center and plasma cell formation [
22], germinal center B cells interact with ICOS+ follicular helper T (Tfh) cells [
23], and antibody-secreting plasma B cells interact with activated Th cells in periarteriolar lymphoid sheaths [
24]. Because the outcome of ICOS signaling in CD4+ T cells depends upon the differentiation and programming of T cells in each context [
25], these interactions represent potentially distinct points of control for antibody responses, where modulation of ICOS signaling by regulated B7h expression on antigen presenting cells could have different effects.
Previous in vitro studies have shown that down-regulation of B7h expression on B cells after activation can restrict ICOS costimulation in cognate CD4+ T cells, suggesting that regulation of B7h levels on activated B cells in vivo could be a control mechanism in T cell-dependent antibody responses. B7h expression on activated B cells is transcriptionally extinguished by exposure to antigen and IL-4 [
26], and is also limited by rapid ectodomain shedding induced either by binding to ICOS or by antigen receptor signaling [
27]. Conversely, B7h expression can be restored by the reactivation of transcription induced by CD40 signaling [
26], and enhanced by the inhibition of ectodomain shedding induced by TLR7/8 and TLR9 signaling [
27]. Thus, the level of B7h expression on activated B cells reflects the integration of multiple critical B cell signaling pathways.
To investigate the cell type-specific functions of regulated B7h expression on B cells and DCs in vivo, we generated transgenic mice with lineage-restricted B7h expression. Our results show that ICOS costimulation mediated by DCs contributes to germinal center formation in response to T cell-dependent antigens and leads to a selective increase in antibody class switching to the IgG2a isotype, without affecting the overall magnitude of antibody responses. In contrast, increased expression of B7h on B cells, but not on dendritic cells, markedly enhances both the number of plasma cells secreting antigen-specific high affinity class-switched antibodies and the serum concentration of such antibodies, without affecting the number of germinal center B cells or antibody isotype usage. Thus, we have defined distinct roles of ICOS costimulation mediated by DCs and plasma B cells in T cell-dependent antibody responses.
Discussion
Analyses of both ICOS
−/− and B7h
−/− mice have demonstrated the critical requirement of ICOS costimulation for normal germinal center formation, antibody isotype switching, affinity maturation, and antibody production in T cell-dependent antibody responses [
1,
7,
13,
14]. However, knockout mice exhibit defects at multiple stages of T cell-dependent antibody responses, ICOS costimulation can have diverse effects on immune responses depending on the context of signaling, and B7h is expressed on multiple antigen presenting cell types involved in antibody responses. Thus, studies of knockout mice could not dissect the potential cell type-specific roles of B7h expression on antigen presenting cells in T cell-dependent antibody responses. We have extended previous studies using transgenic mice with lineage-restricted B7h expression to define selective roles for ICOS costimulation mediated by DCs and plasma B cells in the context of antibody responses.
Our results demonstrate that endogenous B7h expression is high on naïve and germinal center B cells, but is reduced on activated and plasma B cells, and transgenic induction of B7h expression on activated and plasma B cells in B-B7hTg mice enhances the number of antibody-secreting cells produced in response to T cell-dependent antigens. The observed expression pattern of the B-B7h transgene may most likely be explained as a result of the use of an immunoglobulin promoter in place of the endogenous B7h promoter sequences to drive transgene expression, as the immunoglobulin promoter is upregulated in plasma B cells, and possibly also by insertion of the transgene at a genomic location that is available for transcription only in activated B cells. The latter possibility could potentially be problematic if the transgene insertion disrupted an endogenous gene that is important for B cell function. Fortunately, we did not observe any defects in B cell function in B-B7hTg mice in vitro or in vivo, leading us to conclude that such a disruption is unlikely to have occurred in this case.
While we observed induction of B7h expression in B-B7hTg B cells after BCR cross-linking in vitro, we were unable to specifically identify analogous recently activated B cells in vivo in our system, as the frequency of NP-reactive B cells in naïve mice is very low, and the low affinity of primary naïve B cells for NP precluded direct identification of NP-specific cells by FACS based on NP binding in the first few days after immunization in our hands. Although we could not directly measure B7h expression levels on recently activated B cells in vivo, our in vitro results suggest that the B-B7h transgene may lead to overexpression of B7h on these cells in the first 2–3 days, a time frame in which they interact with cognate activated T cells in interfolicular regions [
34,
35]. Thus, it is possible that increased ICOS costimulation in T cells interacting with responding follicular B cells in the first 2–3 days after immunization contributed to the increased antibody responses in B-B7hTg mice. However, it is not clear how ICOS signaling in early cognate B-T interactions could augment the levels of both pre-germinal center low affinity and post-germinal center high affinity antibody production without affecting the number of germinal center B cells.
In addition to expression on in vitro-activated B cells, the B-B7h transgene consistently led to dramatic overexpression of B7h on responding plasma B cells in vivo, and an expansion of these cells in comparison to non-transgenic control mice. The correlation between B7h overexpression on plasma cells and cellular expansion of plasma cells in B-B7hTg mice suggests that differentiated plasma B cells make contact with ICOS-expressing activated T cells, and this interaction regulates the magnitude of antibody responses based in part on the level of B7h expression on B cells, which influences the strength of ICOS signaling in cognate T cells. This putative plasma cell-T cell interaction could potentially occur between Tfh cells and newly differentiated plasma cells in germinal centers, and/or between activated Th cells and plasma cells that co-localize in peri-arteriolar lymphoid sheaths (PALS) in lymph nodes and spleen [
24]. The observed increase in the number of antibody-secreting cells in B-B7hTg mice at day 7 implies that plasma cell-T cell interactions that regulate antibody responses can likely occur outside of germinal centers, as the majority of antibody secreting cells at this early time point are not germinal-center derived [
33].
ICOS costimulation in T cells is dynamically regulated on multiple levels. A previous study generated transgenic mice that overexpress ICOS on all T cells [
36], and made the counterintuitive finding that antibody and germinal center responses were reduced in ICOS transgenic mice, as a result of the increased clipping of B7h from the cell surface of APCs after contact with ICOS [
27]. The resulting reduction of B7h expression on APCs in ICOS transgenic mice caused a reduction of ICOS costimulation in vivo compared with WT mice, leading to defects in B cell responses to T cell-dependent antigens. In addition to ICOS contact, previous in vitro studies have shown that IL4-R and antigen receptor signals reduce cell surface B7h levels on B cells through transcriptional and post-transcriptional mechanisms, while the addition of CD40, TLR7/8 and/or TLR9 signals can limit the extent of B7h downregulation after activation or ICOS contact, causing a relative increase in ICOS co-stimulation in cognate T cells [
26,
27]. In our B-B7hTg model the B7h transgene was induced on activated B cells as a result of the transgenic promoter, leading to an increase in cell surface B7h levels on activated B cells, which was most notable in vivo on differentiated plasma B cells. Ultimately this increased cell surface B7h expression led to expansion of plasma cells in vivo, presumably through increased ICOS co-stimulation in cognate T cells.
These results suggest that B cells regulate the magnitude of antibody responses in part by modulating B7h levels in response to multiple signaling pathways. In this model, physiological settings where endogenous B7h expression on B cells is enhanced or stabilized during antibody responses, such as by CD40 or TLR9 signaling, are predicted to lead to increased class-switched antibody production. Conversely, the results of studies of ICOS transgenic mice suggest that stimuli that lead to an increase in ICOS expression on T cells may counter-intuitively reduce antibody responses through increased clipping of B7h from the cell surface of cognate APCs.
Our results have potential practical implications for the improvement of vaccination strategies, as they demonstrate that ICOS signaling is not only required for many aspects of T cell-dependent antibody responses, but that the strength of ICOS signaling, which is controlled in part by changing levels of B7h on APCs, can modulate antibody responses in the right context. Hence, vaccination strategies that increase ICOS costimulation either directly with ICOS agonists, or indirectly by attempting to stabilize B7h expression on B cells through CD40 or TLR signaling, may be useful to augment the magnitude of vaccine-induced high affinity antibody responses. Conversely, ICOS blockade may be a useful strategy for reducing the expansion of plasma cells secreting pathogenic antibodies in the setting of B cell-mediated autoimmunity [
37,
38]. Interactions between plasma B cells and ICOS+ T cells in the kidneys of lupus patients have been correlated with exaggerated plasma cell generation and pathogenic antibody production [
39]. Our data suggest that the level of ICOS signaling in such interactions can regulate the plasma cell response, and blocking ICOS costimulation in this setting may be a useful therapeutic strategy to reduce the production of auto-antibodies.
Our results show that overexpression of B7h on activated and plasma B cells led to an increase in high-affinity IgG production 21 days after immunization (Figure
5). In this experiment, although a statistically significant difference in high-affinity NP
1-binding IgG could be detected between B7h
+/+ and B7h
−/− mice, the difference in total NP
15-binding IgG was not significant. This was surprising, as numerous studies have reported defects in class-switched antibody production in B7h
−/− and ICOS
−/− mice or after ICOS blockade in WT mice [
7,
14,
15,
40]. However, several studies showed a decrease in IgG1, but not IgG2a production, after primary immunization [
7,
40], suggesting that all isotypes may not be equivalently affected, and the extent of the observed differences vary between studies using different immunogens, adjuvants, and timing of sample collection [
14]. In the case of the experiment shown in Figure
5, we may have failed to detect a difference between B7h
+/+ and B7h
−/− responses because we measured all IgG isotypes together, and took blood samples at day 21 for the purpose of measuring affinity maturation, in contrast to previous reports that show differences specifically for the IgG1 isotype, at 7 and 14 days after immunization. In fact, in a similar experiment measuring total NP
15-binding IgG at day 14, we did detect a statistically significant difference between B7h
+/+ and B7h
−/− animals (Figure
6B), suggesting that while IgG responses in B7h
−/− mice are initially deficient, at later time points total antigen-specific IgG in immunized B7h
−/− mice may reach wild-type levels, although high-affinity (NP
1-binding) antibody responses were deficient as a result of limited germinal center activity in knockout mice.
Initial studies reported Th1 polarization [
16], reduced IL-4 production [
15,
41], and inhibition of Th2-mediated lung mucosal inflammation [
40,
42] in the absence of ICOS signaling, suggesting a role for ICOS in Th2 polarization. Additional studies have paradoxically shown amelioration of Th1-mediated allograft rejection [
43], EAE [
25,
44], and diabetes [
45] in the absence of ICOS signaling, as well as enhancement of Th1-driven IgG2a production by an activatory B7h-Ig fusion protein [
9]. Our results suggest these observations are not contradictory, but rather the outcome of ICOS signaling in activated T cells is highly dependent on the context of B7h-ICOS interactions, as B7h expressed on dendritic cells, but not plasma B cells, can specifically enhance Th1-driven IgG2a responses, while B7h expression on plasma B cells increases IgG production of all isotypes. This may in part reflect the fact that responding DCs and plasma B cells can localize to separate micro-environments, and potentially interact with CD4+ T cells in different states of differentiation that regulate distinct aspects of antibody responses.
In addition to the effect on antibody isotype switching, we show that B7h expression on dendritic cells also contributes to germinal center formation, which leads to antibody affinity maturation. However, the number of responding germinal center B cells in mice that only express B7h on DCs (DC-B7hTg x B7h
−/−) was reduced in comparison to wild-type B7h
+/+ mice (Figure
5b), suggesting that additional ICOS costimulation mediated by other cell types is necessary for optimal germinal center responses and antibody affinity maturation. Since ICOS expression is high on Tfh cells in germinal centers and endogenous B7h expression is high on germinal center B cells, it is likely that germinal center B cell-mediated ICOS signaling in Tfh cells also contributes to the size and dynamics of germinal centers and the extent of antibody affinity maturation in wild-type mice.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
SB generated the DNA construct used to create DC-B7hTg mice. LL generated and initially characterized the B-B7hTg mouse line. KL generated the DC-B7hTg mouse line, performed all experiments with results shown, and drafted the manuscript. BS contributed to the study design and supervised the studies. All authors read and approved the final manuscript.