Identification of costimulatory molecule ICOS expression on human ILC3s

To investigate ICOS expression on ILC subgroups, flow cytometry analysis combined with t-distributed stochastic neighbor embedding (t-SNE) analysis of lymphocytes isolated from noninflamed palatine tonsils was applied; the results showed that ILC3s were the predominant population within tonsillar ILCs (Fig. 1A–D). Despite containing distinctive clusters, tonsillar ILCs (lineage− CD127+) shared some characteristics with tonsillar T cells, including partial expression of ICOS and CD69 as well as low expression of CD40L (Fig. 1A). In contrast, the natural cytotoxicity receptor (NCR) NKp44 was expressed on ILCs but not on T cells (Fig. 1A). Among the three defined ILC subgroups, ICOS expression was detectable on ILC3s (lineage− CRTH2− CD117+), some ILC1s (lineage− CRTH2− CD117−), and some cells resembling ILC2s (lineage− CRTH2+ CD117−) as previously reported [11] (Fig. 1B, Supplementary Fig. 1a–c). Additionally, ICOShigh tonsillar lymphocytes contained not only CD3+ and CD4+ cells but also a certain population of ILC3-like cells (Fig. 1E); furthermore, the percentage of ICOS+ ILC3s in the tonsil and lung was generally higher than that in peripheral blood (PB) (Fig. 1F), and this higher expression was accompanied by the expression of the ILC3 activation-related biomarker NKp44 [18, 19] (Supplementary Fig. 1d).

Fig. 1

ICOS expression is not limited to T cells and ILC2s. A, B Corresponding to the gating strategy in Supplementary Fig. 1a in detail; t-distributed stochastic neighbor embedding (t-SNE) analysis was performed using live tonsillar CD45+ cells. A Using identified biomarker sets, the T-cell cluster was manually gated as CD3+ cells (black), and the cluster of total innate lymphoid cells (ILCs) was identified as lineage (CD3 CD19 CD20 CD14 CD94 CD34 CD1a CD11c CD123 TCRα/β TCRγ/δ FcεRIα)− CD127+ (red). B In ILCs, a second t-SNE analysis was conducted to focus on three clusters corresponding to ILC1s (green), ILC2s (blue), and ILC3s (red). Normalized intensities of each biomarker were calculated and displayed on the plot. The results correspond to three donors. Frequencies (C) and cell counts (per gram) (D) of total ILCs (lineage− CD127+), ILC1s (lineage− CD127+ CRTH2− CD117−), ILC2s (lineage− CD127+ CRTH2+), and ILC3s lineage− CD127+ CRTH2− CD117+) from gated sets of live tonsillar CD45+ cells are shown in FACS dot plots (n = 11). E The inducible T-cell costimulator (ICOS)high lymphocytes were gated and indicated as CD3− or CD4− ILC3s. F ICOS expression of ILC3s (left) and frequencies of ICOS+ ILC3s in ILC3s from human tonsil, distal lung sample (more than 5 cm from the tumor lesion) from donors with lung cancer, and peripheral blood of healthy donors (n = 5–11). G Coexpression of ICOS and common biomarkers on tonsillar ILC3s. The error bars indicate the mean ± standard error of the mean (SEM)

Some known activation-associated molecules were detected on ICOS+ ILC3s but to a lesser extent on ICOS− ILC3s (Fig. 1G). However, the expression of CD28 and the conventional checkpoint PD-1 was low on tonsillar ILC3s, even on ICOS+ ILC3s (Fig. 1G). In contrast to ILC2s [11], ILC3s lacked ICOSL expression (Fig. 1G). Collectively, the costimulatory molecule ICOS may play a more indispensable role in ILC3 biology than in T helper cell or ILC2 processes.

ICOS+ ILC3s show more activation-associated and helper-like characteristics

To investigate the effect of ICOS on the characteristics of ILC3s, tonsillar ILC3s, characterized by RORγt (Supplementary Fig. 2a, b), were isolated and then activated in vitro. When stimulated by IL-7 + IL-2 + IL-1β + IL-23, the ICOS level on ILC3s further increased, and ILC3s were completely converted to ICOS+ ILC3s on Day 4. In contrast, despite some increase in expression, the degree of ICOS upregulation was not as remarkable after IL-7 + IL-2 stimulation as that after IL-7 + IL-2 + IL-1β + IL-23 stimulation (Fig. 2A). This result was confirmed using PB-derived ILC3s (Supplementary Fig. 2c). In line with this, ILC3s were enlarged after IL-7 + IL-2 + IL-1β + IL-23 stimulation, with inflated nuclei and cytoplasm and several vesicles within the cytoplasm (Fig. 2B), which were accompanied by enhanced ILC3 survival, activation, and cytokine (IL-22, IL-17A, TNF, IFN-γ, and GM-CSF) production (Fig. 2C, Supplementary Fig. 2d, e). These findings suggested that ICOS may act as an activation-associated molecule on the ILC3 surface.

Fig. 2

ICOS+ ILC3s are more like helper cells than ICOS− ILC3s based on their transcription profile and functions. A–C Purified ILC3s from the same donor were stimulated by interleukin (IL)-2 + IL-7 or IL-2 + IL-7 + IL-1β + IL-23. A Median fluorescence intensity (MFI) for ICOS on ILC3s. B Morphological comparison of ILC3s (white arrows indicate vesicles) on Day 7. C Representative flow plots and quantification of IL-22+ ILC3s (n = 4). D Volcano plot comparison (left) of differentially expressed genes (DEGs) between tonsillar ICOS− and ICOS+ ILC3s. The heatmap (right) shows the DEGs related to the immune system and environmental information processing. DEGseq analysis (log 2 of fold change ≥ 1.5, Q ≤ 0.005) was used to identify the DEGs (n = 3). The results are shown as a scatter plot of the Z score (row direction) for fragments per kilobase million mapped (FPKM). E–H Purified tonsillar ICOS− and ICOS+ ILC3s were treated with IL-7 + IL-2 + IL-1β + IL-23 and analyzed on Day 4 and/or 7. E Representative fluorescence-activated cell sorting (FACS) profiles for ICOS− ILC3s (light blue) and ICOS+ ILC3s (light red) overlapped by ICOS and NKp44 expression at Day 0, 4, and 7. F Change in relative ICOS expression on ICOS− and ICOS+ ILC3s on Day 0, 4, and 7 (n = 3). G, H Flow cytometry profiles and summary of the frequencies of viable (G upper), ki-67+ ILC3s (G upper-middle), NKp44+ ILC3s (G lower-middle), and CD69+ ILC3s (G lower) as well as IL-22+ ILC3s (H upper) and IL-17A+ ILC3s (H lower) treated with IL-7 + IL-2 + IL-1β + IL-23 on Day 4 (n = 4–10). I Morphological comparison of ILC3s vs. CD4+ T cells sorted from tonsillar lymphocytes. J Representative FACS profiles for ICOS expression of tonsillar CD4+ (left) and CD8+ T cells (right). K Sorted tonsillar ICOS−/+ ILC3s and ICOS−/+ CD4+ T cells were incubated at the same cell concentration and stimulated by IL-7 + IL-2 + IL-1β + IL-23 (for ILC3s) or anti-CD3/CD28 antibody (for CD4+ T cells) in vitro. On Day 4, IL-22, IL-17A, total TNF, GM-CSF, and IFN-γ levels in the supernatants were determined using enzyme-linked immunosorbent assay (ELISA) and cytometric bead array (CBA) (J). The results generated from different donors (n = 4–8) were combined. The error bars indicate the mean ± SEM. Statistical significance was determined using Student’s paired t-test (C, F–H) or unmatched one-way analysis of variance (ANOVA) and Tukey’s multiple comparison test (K)

To further elucidate the role of ICOS in ILC3s, tonsillar ILC3s were divided into ICOS− and ICOS+ ILC3s and sorted for RNA sequencing analysis (Fig. 2D, Supplementary Fig. 3a, b, Supplementary Table 1). Interestingly, ICOS+ ILC3s shared some genes with activated ILCs (NCR1, NCR2, CD160, and IL13) [20, 21] and T helper cells (IL2, IL22, CSF3, TIGIT, CTLA4, PDCD1, TNFSF14, and TNFSF13B). However, ICOS− ILC3s also had the expression of some genes involved in antigen presentation, such as the MHCII family genes [22] (HLA-DQB1, HLA-DOA, and CD74) and B-cell-associated genes (CD86, BTK, CD22, TNFRSF13B, and TNFRSF19) [23]. Next, ICOS− and ICOS+ ILC3s were stimulated by IL-7 + IL-2 + IL-1β + IL-23 in vitro. Although ICOS expression increased on ICOS− and ICOS+ ILC3s, the ICOS levels on ICOS+ ILC3s were higher than those on autologous ICOS− ILC3s and peaked on Day 4 after activation (Fig. 2E, F). In line with the increased ICOS expression, the proportions of viable ILC3s, ki-67+ ILC3s, NKp44+ ILC3s, and CD69+ ILC3s in ICOS+ ILC3s were higher than those in ICOS− ILC3s, and the same trend was observed for intracellular IL-22 and IL-17A expression (Fig. 2G, H). These observations indicated that ICOS might act as an activation-associated molecule rather than a specific biomarker of ILC3s.

In line with the ICOS expression results, ILC3s and CD4+ T cells showed remarkably similar cell size and morphology and a similar nuclear-cytoplasmic ratio, with oval- or bean-shaped single nuclei and oligoplasmic lymphoid morphology (Fig. 2I, J). To investigate their functional heterogeneity, ICOS−/+ ILC3s and ICOS−/+ CD4+ T cells were cultured at the same cell count and concentration. The results showed that the levels of cytokines secreted by ICOS+ ILC3s, especially that of IL-22, were higher than those secreted by ICOS− ILC3s at Day 4 (Fig. 2K) and Day 7 (Supplementary Fig. 4a). The levels of some cytokines, such as IL-22, TNF, IFN-γ, and GM-CSF, produced by ICOS+ ILC3s were higher than those produced by ICOS+ CD4+ T cells. In contrast, the IL-17A-secreting capacity of ICOS+ ILC3s was lower than that of ICOS+ CD4+ T cells. These results indicated that tonsil-resident ICOS+ ILC3s might play a unique role in the immune environment via specific secretion of cytokines even though there are fewer ICOS+ ILC3s than T helper cells.

ICOS costimulation promotes the production of IL-10 by regulatory T cells (Tregs) [8], and ILCs contain a subgroup of IL-10+ ILCs, which are known as regulatory ILCs (ILCregs) [24]. Thus, we investigated whether ICOS+ ILC3s have the characteristics of ILCregs. Neither ICOS− ILC3s nor ICOS+ ILC3s had the ability to secrete IL-10; however, both ICOS- CD4+ T and ICOS+ CD4+ T cells secreted some IL-10 (Supplementary Fig. 4b), in agreement with previous studies [25, 26], suggesting that ICOS+ ILC3s and ILCregs [24] are two distinct and nonoverlapping ILC subpopulations. Collectively, the role of ICOS in ILC3s is heterogeneous, distinct from that in ICOS+ T helper cells, and cell type-specific.

ICOS costimulation enhances ILC3 functions

To elucidate the influence of the ICOS/ICOSL interaction on human ILC3s, we used ICOS costimulation assays as previously reported [27, 28]. The results showed that costimulation of ILC3s with precoated human recombinant soluble ICOSL protein (rsICOSL) enhanced the survival, proliferation, and activation of ILC3s, as indicated by an increased proportion of viable, ki-67+ and CD69+ ILC3s (Fig. 3A), even without NKp44+ and CD25+ ILC3 expansion (Supplementary Fig. 5a). In addition, the concentrations of IL-22, IL-17A, IFN-γ, TNF, and GM-CSF were significantly upregulated after ICOSL engagement (Fig. 3B). However, free rsICOSL failed to enhance cytokine production by ILC3s (Supplementary Fig. 5b), which indicated that the effect of ICOS costimulation on ILC3s primarily depended on ILC3 cell-to-cell contact via surface molecules with the ICOSL provider rather than via interaction with free ICOSL in the microenvironment.

Fig. 3

ICOS costimulation enhances the survival, proliferation, and activation of ILC3s. A, B Tonsillar ILC3s were incubated in human recombinant soluble ICOSL protein (rsICOSL)-precoated or uncoated wells (control) in the presence of IL-7, IL-1β, and IL-23 for 4 days. A Representative flow plots and quantification of viable ILC3s (upper), ki-67+ ILC3s (middle), and CD69+ ILC3s (lower). The numbers in the quadrant represent the percentages of cells (n = 5–9). B Cytokines in the supernatants at Day 4 as determined using ELISA and CBA (n = 9–14). C Histogram comparing ICOSL expression among the CD32/ICOSL-expressing cell line (red line) and CD32-expressing cell line (black line), with the isotype control shown in gray. D Purified tonsillar ILC3s were cocultured with the parental CD32 line (control) and CD32/ICOSL line in the presence of IL-7, IL-1β, and IL-23 for 4 days. Cytokine levels in the supernatants were determined using ELISA and CBA (n = 6–10) on Day 4. The error bars indicate the mean ± SEM, and P values were determined using a two-tailed Student’s paired t-test

To verify the immunological function of ICOSL+ cells on ILC3s, a human cell line expressing full-length ICOSL (CD32/ICOSL-expressing cells) was generated as previously described [26] (Fig. 3C) and cocultured with tonsillar ILC3s. Consistent with the effect of precoated rsICOSL Fig. 3B, the CD32/ICOSL-expressing cell line also increased the production of IL-22, IL-17A, IFN-γ, TNF, and GM-CSF by ILC3s (Fig. 3D, Supplementary Fig. 5c). Collectively, the ICOS/ICOSL pathway provides crucial costimulatory signals that reinforce ILC3 activation, and ILC3s are likely to receive activation-modulating signals by interacting with ICOSL+ cells in the immune microenvironment.

Bidirectional promotion between tonsillar ILC3s and autologous B cells

B cells are not only the main provider of ICOSL [9, 29] but also the predominant immune cells in the tonsil (Supplementary Fig. 6a). Resembling T cells, RORγt+ CD3− ILC3s also colocalized with B cells in vivo (Fig. 4A). To study the bidirectional effects between ILC3s and B cells, autologous tonsillar ILC3s and B cells were sorted and cocultured at different ratios (Fig. 4B). The results showed that ILC3s and B cells promoted the survival of each other, and the survival of these cells was ratio-dependent (Supplementary Fig. 5b).

Fig. 4

ILC3s and B cells in coculture promote the specific functions of each other. A Immunofluorescence-based histological analysis of tonsils. ILC3s (ROR-γt+ CD3−) and B cells (CD20−) were in close contact with each other in vivo. Solid arrows, ILC3s; open arrows, B cells; *, T cells. Scale bar = 20 μm. B–I Tonsillar ILC3s and B cells (CD3− CD19+) sorted from the same donor were cocultured for 4 days and subjected to FACS analysis (B), and ELISA and CBA were used to determine the phenotypes and functions of ILC3s (CD45+ CD19− CD20− ICOS+) and B cells (CD45+ CD19−/+ CD20+ ICOS−). C–E B cells were cocultured with autologous ILC3s at a 5:1 ratio in the presence or absence of ODN2006 for 4 days. C Representative flow plots and quantification of viable (upper) and ki-67+ (lower) B cells (n = 4–5). D MFI and quantification of ICOSL on B cells incubated for 4 days with or without ILC3s (n = 5). E ELISA and CBA for IgA, IgM, and total IgG in the supernatants of B cells incubated for 4 days with or without ILC3s (n = 4–9). F–I ILC3s were cocultured with autologous B cells at a 1:15 ratio in the presence of IL-7 + IL-2 or IL-7 + IL-2 + IL-1β + IL-23 for 4 days. F Representative flow plots and quantification of viable (upper), ki-67+ (middle), and NKp44+ ILC3s (down) (n = 5–7). G ELISA and CBA for IL-22, IL-17A, total TNF, GM-CSF, and IFN-γ in the supernatants of ILC3s incubated for 4 days with or without B cells (n = 5–9). H, I ELISA for IL-10 in the supernatant of ILC3s incubated for 4 days with or without B cells (H) and ICOSL protein (I). The results are representative of five (H) or seven (I) experiments. The error bars indicate the mean ± SEM. J Representative flow plots of IL-10+ ILC3s and IL-10+ B cells after coculture for 4 days. Statistical significance was determined using Student’s paired t-test (C–H) or matched one-way ANOVA and Tukey’s multiple comparison test (I)

Regarding the influence of ILC3s on B cells, ILC3s promoted the survival and proliferation of B cells, particularly in the presence of oligonucleotide (ODN) 2006 (a class B CpG ODN that can be used for B-cell activation as a ligand of TLR9, Fig. 4C), and the level of detectable ICOSL on the surface of B cells was reduced (Fig. 4D), suggesting that ICOSL was shed due to interactions with ICOS+ ILC3s [30]. Because IL-1β induces the production of IgA, IgM, and IgG by B cells (Supplementary Fig. 7a), in this ILC3-B-cell coculture system, IL-1β was excluded. Functionally, ILC3s promoted the secretion of IgA and IgM, but not IgG, by B cells; however, this impact was not remarkable in the absence of ODN2006 stimulation (Fig. 4E). As the ICOSL level is downregulated after interaction with ICOS [31], the ICOS signaling pathway may participate in the process of B-cell priming by ILC3s.

Regarding the effect of B cells on ILC3s, B cells were cocultured with ILC3s in the presence of IL-7 + IL-2 or IL-7 + IL-2 + IL-1β + IL-23 stimuli. We found that after coculture with B cells, regardless of the addition of IL-1β and IL-23, the proportion of viable ILC3s, ki-67+ ILC3s, and NKp44+ ILC3s as well as the number of ILC3s increased (Fig. 4F, Supplementary Fig. 7b). CD69 expression on ILC3s was not affected after coculture with B cells (Supplementary Fig. 7c). Moreover, B cells promoted ILC3 production of cytokines, including IL-22, IL-17A, IFN-γ, TNF, and GM-CSF (Fig. 4G, Supplementary Fig. 7d). In the ILC3-B-cell coculture system, ODN2006 was added for B-cell activation; however, despite ICOS+ ILC3s showing expression of TLR9 transcripts (Fig. 2D), ODN2006 did not significantly influence ILC3 activation (Supplementary Fig. 7e, f).

In the supernatant of the ILC3 and B-cell coculture system, we found high IL-10 levels (Fig. 4H), which was consistent with previous findings that ILC3s induce B-cell differentiation into IL-10-producing regulatory B cells (Bregs) [4]. We next determined whether IL-10 was generated by ILC3s. We found that IL-10 was undetectable in the supernatant of ILC3s regardless of stimulation by IL-7 + IL-2 + IL-1β + IL-23 or rsICOSL, but its expression was triggered when B cells were added. (Fig. 4H, I). Furthermore, the IL-10 in the ILC3-B-cell coculture system was mainly derived from B cells and not ILC3s (Fig. 4J). These results suggest that neither activation by IL-7 + IL-2 + IL-1β + IL-23 nor costimulation of ICOS triggers IL-10 production by ILC3s.

Reciprocal promotion of ILC3s and B cells partially requires the ICOS/ICOSL interaction

We next explored whether the reciprocal effects of ILC3s and B cells are contact-dependent or contact-independent. A Transwell assay showed that when cell contact was blocked, the B-cell-induced ILC3-specific production of IL-22, IL-17A, IFN-γ, TNF, and GM-CSF was eliminated (Fig. 5A). Similarly, the ILC3-induced upregulation of IgA and IgM secretion by B cells was inhibited (Fig. 5B). These observations were in line with the decreased IL-10 levels in B cells (Fig. 5C). Although some secreted extracellular factors, such as ILC3-derived BAFF and B-cell-derived IL-15, promote crosstalk between ILC3s and B cells [4, 5], these results indicate that direct contact between ILC3s and B cells is essential for their reciprocal promotion.

Fig. 5

The bidirectional promotion of ILC3s and B cells partially requires ICOS/ICOSL-induced cell‒cell contact. A–C Blocked of tonsillar ILC3-B-cell cell-cell contact in transwell cocultures. ELISA and CBA for IL-22, IL-17A, total TNF, GM-CSF, IFN-γ (A), IgA, IgM (B), and IL-10 (C) in the supernatant of ILC3s and B cells cocultured at a 1:5 ratio for 7 days in the presence of IL-7, IL-23, and ODN2006 with or without a transwell (ILC3s in the upper well, n = 4–6). D–F Effects of ICOS blockade on the ILC3-B-cell interaction. D ELISA and CBA for IL-22, IL-17A, total TNF, GM-CSF, and IFN-γ in the supernatants of ILC3s only (black), cocultured ILC3s and B cells (red), and cocultured ILC3s and B cells with an anti-ICOS antibody (gray) incubated at a 1:15 ratio for 4 days in the presence of IL-7, IL-23, and ODN2006 (n = 5–7). ELISA for IgA, IgM (E), and IL-10 (F) in the supernatants of B cells only (black), cocultured ILC3s and B cells (purple), and cocultured ILC3s and B cells with anti-ICOS antibody (gray) incubated at a 1:5 (E) or 1:15 (F) ratio for 4 days in the presence of IL-7, IL-23, and ODN2006 (n = 4–5). The error bars indicate the mean ± SEM, and statistical significance was determined using Student’s paired t-test (A–C, and F) or matched one-way ANOVA and Tukey’s multiple comparison test (D, E)

To further investigate the role of ICOS signaling in the ILC3-B-cell interaction, we conducted blocking assays using an ICOS-neutralizing antibody followed by functional tests of ILC3s and B cells. ICOS blockade resulted in a partial decrease in IL-22, IL-17A, TNF, GM-CSF, and IFN-γ secretion by ILC3s (Fig. 5D) but little reduction in IgA, IgM, and IL-10 expression in B cells (Fig. 5E, F). These results suggest that ICOS is partially involved in promoting ILC3 activation during the interaction of ILC3s and B cells. Although IL-21 is involved in ICOS-mediated Tfh and B-cell interactions in lymphoid tissues, such as the tonsil [32, 33], neither ICOS- ILC3s nor ICOS+ ILC3s produced IL-21 (Supplementary Fig. 7g). Collectively, the ILC3-B-cell interaction is partially ICOS signaling-dependent, and it may involve other molecules that induce contact between ILC3s and B cells.

B-cell-reinforced ILC3 activation requires the cooperation of ICOS and CD40 signaling

The costimulatory molecule CD40 generally supports the ICOS/ICOSL interaction in T-cell-B-cell responses [9, 34]. Thus, we investigated the role of CD40 signaling in the ILC3-B-cell interaction. Of note, CD40L expression was not detected on fresh tonsil-derived, lung-derived, or PB-derived ILC3s (Fig. 6A) but appeared on the ILC3 surface after activation by IL-7 + IL-2 + IL-1β + IL-23 (Fig. 6B). The percentage of induced CD40L+ ILC3s was further increased after activation (Fig. 6C). Moreover, ICOS+ ILC3s had a stronger potential for CD40L expression, and the percentage of CD40L+ ILC3s increased after ICOSL engagement (Fig. 6D, E). During Tfh-B-cell responses, the ICOS/ICOSL interaction promotes the expression of CD40L on T cells [9, 34, 35], suggesting that ILC3s may function similarly to Tfh cells during interaction with B cells, and the effect of CD40 likely occurs later than the ICOS/ICOSL interaction.

Fig. 6

B-cell-induced activation of ILC3s requires the cooperation of ICOS and CD40 signaling. A Representative flow plots for CD40L and ICOS expression in ILC3s from human tonsil, distal and normal lung samples from donor with lung cancer, PB of healthy donor, and CD4+ T cells from human tonsil. B, C Representative flow plots and quantification of CD40L+ ILC3s in tonsillar ILC3s stimulated by IL-2 + IL-7 + IL-1β + IL-23 for 1, 4, and 7 days (B) or incubated with or without IL-1β + IL-23 for 4 days (C). The results are representative of three donors in (B) and seven experiments in (C). D Representative flow plots and quantification of CD40L+ ILC3s in ICOS− and ICOS+ ILC3s stimulated by IL-2 + IL-7 + IL-1β + IL-23 for 4 days (n = 10). E Representative flow plots and quantification of CD40L+ ILC3s in purified tonsillar ILC3s incubated with or without ICOSL protein in the presence of IL-7, IL-1β, and IL-23 for 4 days (n = 5). F Representative flow plots and quantification of CD40L+ ILC3s in tonsillar ILC3s cocultured with autologous B cells at a 1:15 ratio in the presence of IL-7 + IL-2 or IL-7 + IL-2 + IL-1β + IL-23 for 4 days (n = 5). The effects of CD40 blockade on the ILC3-B-cell interaction were assessed (G–I). ELISA of IgA, IgM (G), and IL-10 (I) in the supernatants of B cells only (black), cocultured ILC3s and B cells (purple), and cocultured ILC3s and B cells with anti-CD40 antibody (gray) incubated at 1:5 (E) or 1:15 (F) ratios for 4 days in the presence of IL-7, IL-23, and ODN2006 (n = 6). I ELISA and CBA for IL-22, IL-17A, total TNF, GM-CSF, and IFN-γ in the supernatants of ILC3s only (black), cocultured ILC3s and B cells (red), and cocultured ILC3s and B cells with anti-CD40 antibody (gray) incubated at a 1:15 ratio for 4 days in the presence of IL-7, IL-23, and ODN2006 (n = 5–7). J, K Tonsillar ILC3s were cocultured with autologous B cells in the absence (blank column) or presence of a different concentrations of anti-CD40 antibody (black column) plus the ideal concentration of anti-ICOS antibody (gray column) for 4 days. J ELISA of IL-22 in the supernatant of the ILC3-B-cell coculture system (ratio = 1:15). K CBA of IgA and IgM in the supernatant of the ILC3 and B-cell coculture system (1:5). The results are representative of five (J) or six (K) experiments. The error bars indicate the mean ± SEM. *P < 0.05, **P < 0.005, ***P < 0.0005, ****P < 0.0001, and NS P ≥ 0.05. Statistical significance was determined using Student’s paired t-test (C–F, H, J, and K) or matched one-way ANOVA and Tukey’s multiple comparison test (G, I)

To determine the role of CD40 signaling in the ILC3-B-cell interaction, we used a CD40-neutralizing antibody in the ILC3 and B-cell coculture system to perform a CD40 blockade assay. The results showed that once ILC3s interacted with B cells, CD40L on ILC3s was undetectable (Fig. 6F), and CD40 blockade inhibited the ILC3-induced upregulation of IgA, IgM, and IL-10 levels in B cells (Fig. 6G, H). In addition, the B-cell-induced increase in ILC3 cytokine secretion was partially inhibited after CD40 blockade (Fig. 6I), suggesting that CD40 signaling mediated the interaction between the two cells.

To further investigate the relationship between ICOS and CD40 signaling in the ILC3-B-cell interaction, we prepared anti-ICOS and anti-CD40 antibodies in gradient concentrations. Regarding ILC3s, the inhibitory effect of anti-CD40 antibody on ILC3-produced IL-22 was dose-dependent in the absence of anti-ICOS antibody. In the presence of an optimal dose of anti-ICOS antibody plus an altered concentration of the anti-CD40 antibody, IL-22 was inhibited much more significantly; however, the dose-dependence of the anti-CD40 antibody was partially weakened (Fig. 6J). In B cells, IgA and IgM expression was inhibited after CD40 blockade in a dose-dependent manner in the absence of anti-ICOS antibody. The anti-ICOS antibody acted synergistically with the anti-CD40 antibody; however, the inhibitory function of the anti-ICOS antibody was concealed in the presence of abundant anti-CD40 antibody (Fig. 6K). In summary, ICOS and CD40 signaling play a synergistic role in the B-cell-mediated promotion of ILC3s, with ICOS signaling acting as a major component. In addition, ILC3-induced enhancement of B-cell functions is primarily dependent on CD40 signaling.