It has been previously shown that β-glucan particle size significantly affects inflammatory cytokine secretion from myeloid-derived cells with smaller particulate forms of β-glucan found to be weak inducers of cytokine responses. This study aimed to characterize further the differences in cytokine responses elicited by either the large particulate β-glucan, curdlan (up to 0.2 mm diameter), or β-glucan-microparticles (glucan-mp; 1–5 µm diameter) (14), in human mDC by measuring IL-1β, IL-6, IL-23, TSLP, and CCL22 secretion in response to these different stimuli. Characterization of both curdlan and glucan-mp by high field 1H-NMR showed these β-glucans to have very similar primary structures (Supplementary Image 1). mDC stimulated with curdlan produced large amounts of IL-1β, IL-6, and IL-23, as compared to glucan-mp stimulated mDC secreted much reduced, though still detectable, quantities of IL-1β, IL-6, and IL-23 (Figures 1A–C). In marked contrast, curdlan- and glucan-mp induced secretion of similar quantities of TSLP and CCL22 (Figures 1D,E). However, since curdlan and the glucan-mp were derived from different microorganisms, A. faecalis (a Gram-negative bacterium) and the yeast S. cerevisiae, respectively, we wished to ensure that the findings were not due to differences in the species from which they were prepared. Therefore, we produced curdlan microparticles (curdlan-mp), using a previously described sonication protocol (14). Once more, we observed much reduced amounts of IL-1β, IL-6, and IL-23 in response to curdlan-mp as compared to curdlan, but no difference in TSLP or CCL22 (Figures 1F–J). Additionally, mDC stimulated with curdlan-or glucan-mp displayed no difference in activation status, as similar levels of surface expression of CD11c, CD86, and HLA-DR were observed (Figure 2). Thus, the results were consistent with a difference in the size of the different β-glucan preparations affecting cytokine production. To examine whether the effects observed were due to differences in gene transcription, we also tested whether particle size resulted in similar differences in cytokine gene transcription and found that IL-1β, IL-6, IL-23p19, and TSLP mRNA expression mirrored the data on secretion of each cytokine (Figures 3A–H). Thus, small particles induce reduced inflammatory gene transcription compared to larger particles.

Our data suggested that small particulate β-glucans are not invariably poor inducers of cytokine secretion, but that certain cytokines are more sensitive to the size of the β-glucan particle used. Previous work has highlighted the importance of IL-1β secretion in orchestrating dectin-1-mediated immune responses (12, 16, 23). Therefore, we examined whether the amount of IL-1β stimulated by β-glucan particles of different sizes is critical in determining the amounts of other cytokines produced. We showed that pro-IL-1β protein expression in response to glucan-mp-stimulation was markedly reduced compared to curdlan stimulation (Figure 4A). Furthermore, exogenous recombinant IL-1β added at similar concentrations to that produced by curdlan stimulation recovered glucan-mp IL-6 and IL-23 secretion to levels equivalent to those produced with curdlan stimulation (Figures 4B,C). In contrast, amounts of CCL22 secreted were unaffected by addition of exogenous IL-1β (Figure 4E), while TSLP secretion was augmented, but only approximately twofold (Figure 4D), as compared to the >10-fold increase seen for IL-6 and IL-23. This result is in agreement with our previous finding that there is a requirement for signaling through the IL-1 receptor for optimal TSLP production (23). However, it is clear that unlike with IL-6 and IL-23, the additional IL-1β generated with curdlan stimulation of mDC is not essential for TSLP expression as the relatively low amounts of IL-1β produced in response to glucan-mp are sufficient for substantial secretion of TSLP.

Since the amounts of IL-1β produced by mDC in response to large and small β-glucan particles differed substantially, we wished to examine the signaling requirements for the IL-1β production by each form of β-glucan. There are numerous reports showing the importance of generation of ROS, and the induction of hypoxia-inducible factor 1-alpha (HIF-1α), to induce the changes in cellular metabolism required for IL-1β expression during stimulation through dectin-1 (6, 8, 24, 25). We showed that curdlan and glucan-mp induced similar quantities of ROS in mDC (Figure 5A). They also both induced HIF-1α expression, though this was more marked with curdlan (Figure 5B). To determine the contribution of NADPH-derived ROS (termed ROS from this point) to HIF-1α expression and the induction of IL-1β, we generated mDC from patients with CGD; these patients have mutations in genes encoding components of the NADPH oxidase complex and, therefore, cannot generate ROS. When mDC from CGD patients were stimulated with glucan-mp, neither HIF-1α, nor pro-IL-1β were detected (Figure 5B), indicating that both required NADPH oxidase activity and likely ROS production. In contrast, curdlan stimulation induced pro-IL-1β in mDC from CGD patients, although this was reduced compared to healthy controls, and HIF-1α was not detected, thus indicating a ROS/HIF-1α-independent mechanism of pro-IL-1β production. Accordingly, IL-1β, IL-6, IL-23, and TSLP secretion was effectively abolished in mDC from CGD patients responding to glucan-mp, but not significantly affected when the same cells responded to curdlan (Figures 5C–L). As expected from the findings reported above, normal levels of IL-6 (Figure 5I), IL-23 (Figure 5J), and near normal levels of TSLP (Figure 5K) were produced by mDC from CGD patients responding to curdlan, since sufficient IL-1β was produced even in the absence of ROS and HIF-1α induction. Both curdlan and glucan-mp-induced CCL22 were insensitive to ROS production (Figures 5G,L), with a suggestion of enhanced levels from CGD patients. Thus, these experiments showed that IL-1β production in response to curdlan proceeds in the absence of ROS (and of HIF-1α), whereas this does not occur when glucan-mp are used to stimulate mDC.

As β-glucan size affected the production of IL-1β, and the requirement for NADPH oxidase activity to produce IL-1β, we investigated the mechanism accounting for the effects of particle size. The size of β-glucan particles affects the capacity of murine myeloid-derived cells to phagocytose them and induce inflammatory cytokines (14, 15). Thus, we treated human mDC with cytochalasin D (CYTD) to prevent glucan-mp internalization by phagocytosis and found that this substantially increased the amounts of IL-1β, IL-6, and IL-23 produced in response to glucan-mp, to levels equivalent to those obtained with curdlan stimulation; there was a less marked but significant increase in TSLP production, but no effect on CCL22 secretion since this was already maximal with glucan-mp (Figures 6A–E). Therefore, the inhibition of phagocytosis of glucan-mp had the same effect as the supplementation of glucan-mp-stimulated mDC cultures with exogenous IL-1β. The enhancement of glucan-mp-induced cytokine production by CYTD was independent of NADPH oxidase activity and likely ROS production, since mDC from CGD patients also produced augmented amounts of these cytokines when treated with CYTD (Figures 6F–I). Lack of NADPH oxidase activity had no effect on CCL22 secretion (Figure 6J). To further confirm whether these differences were a result of dectin-1-mediated phagocytosis, we examined the relative capacity of curdlan and glucan-mp to induce a decrease in dectin-1 surface expression on mDC. Glucan-mp-stimulated mDC induced a rapid decrease in surface dectin-1 expression 30 min poststimulation, whereas expression on mDC during curdlan stimulation barely changed in comparison to unstimulated stained controls, and this expression was maintained 4 h post-curdlan stimulation (Figure 7). Moreover, CYTD pretreatment of mDC stimulated with glucan-mp inhibited dectin-1 internalization (Figure 7) and likely prolonged signaling through dectin-1.

It is well established that dectin-1 signaling promotes the differentiation of CD4⁺ T cells to T_H1- and T_H17-cells (3, 16, 26, 27); however, as β-glucan size significantly affected cytokine secretion from mDC, we investigated whether this had an impact on CD4⁺ T cell differentiation. CD3/CD28-stimulated naïve CD4⁺ T cells were cultured in supernatants derived from mDC prestimulated with either curdlan or glucan-mp. CD4⁺ T cells cultured in conditioned media derived from curdlan-stimulated mDC produced significantly higher levels of IL-17 and IFNγ compared to those CD4⁺ T cells cultured in conditioned media from glucan-mp stimulated mDC (Figures 8A,B). This increased production of IL-17 and IFNγ observed in curdlan-treated cells could be replicated when glucan-mp-stimulated mDCs were supplemented with recombinant IL-1β (Figures 8C,D).

Together these data show that failure to phagocytose β-glucan containing particles, as occurs with curdlan, or with glucan-mp in cells treated with CYTD, allows the production of IL-1β in a manner independent of NADPH oxidase activity. However, when phagocytosis of glucan-mp proceeds normally, the small amounts of IL-1β secreted requires NADPH activity, and is, therefore, regulated by an alternative mechanism compared to larger β-glucan particles that are not internalized. Therefore, the small amounts of IL-1β produced in response to internalized glucan-mp are insufficient to allow production of substantial amounts of IL-6 or IL-23 and this is the mechanism by which reduced inflammatory responses to glucan-mp occur. Furthermore, the differences in cytokine milieu generated as a result of β-glucan size modify the capacity to drive T_H1- and T_H17-cell differentiation.

Human blood was sourced from apheresis cones derived from healthy donors (Addenbrooke’s Hospital, Cambridge) and age- and sex-matched CGD patients (Royal Free Hospital, London). Appropriate consent to use blood-derived cells for research was obtained. Ethics Reference Number: 04/Q0501/119.

β-Glucan was isolated from S. cerevisiae using a modification of the method of Williams et al. as described by Mueller and colleagues (29, 30). Specifically, phosphoric acid was used instead of hydrochloric acid in the protic acid digestion step. This approach results in a highly pure glucan, but with less degradation of the glucan polymer structure. The glucan was characterized by 1H-NMR as described by Lowman et al. (17). The glucan (~20 mg) was dissolved in 700 µl of deuterated DMSO_d6. Approximately 50 µl of TFA was added to shift the water peak away from the carbohydrate spectral region. The 1H-NMR spectra were obtained at 80°C. S. cerevisiae glucan was determined to be >95% pure based on the 1D 1H-NMR and comparison to an ultrapure (>97%) glucan reference standard produced in the Williams laboratory.

Human blood was sourced from healthy donors (Addenbrooke’s Hospital, Cambridge) and age, sex-matched CGD patients (Royal Free Hospital, London). Human mDCs were generated from CD14⁺ monocytes isolated from PBMC by magnetic bead separation (Miltenyi) and were differentiated by culturing for 6 days in RPMI1640 (Lonza) supplemented 5% FCS (Biosera), 20 ng/ml GM-CSF (Life Technologies) and 4 ng/ml IL-4 (BD Biosciences) as described previously (23, 31, 32). Naïve human CD4⁺ T cells were isolated by magnetic bead separation (Miltenyi).

Monocyte-derived dendritic cells were stimulated with 50 µg/ml of curdlan (Wako), curdlan microparticles (curdlan-mp) [generated as described previously (14)], and β-1,3 glucan microparticles (glucan-mp) [generated as described above (29, 30)]. mDCs were also stimulated with β-glucans and 2 µM CYTD (Sigma) and/or 10 ng/ml IL-1β (Miltenyi) simultaneously. Cell cultures were incubated for predefined time periods as highlighted in figure legends.

Naïve CD4⁺ T cells were stimulated with anti-CD3/anti-CD28 activation beads (1:1 ratio) (ThermoFisher) and 20 IU of IL-2 (Becton Dickinson) for 7 days in the presence of supernatants derived from mDC prestimulated with either 50 µg/ml of curdlan or glucan-mp. CD4⁺ T cell culture supernatants were subsequently harvested to evaluate IL-17 and IFNγ secretion.

Enzyme-linked immunosorbent assays were performed using appropriate optimized kits for IL-1β, IL-6, IL-23, IL-17, IFNγ (eBioscience), TSLP, and CCL22 (R&D) according to the manufacturers’ instructions. Briefly, plates were precoated with capture antibody overnight and washed five times with wash buffer composed of 1× PBS supplemented with 0.05% Tween 20 (Sigma). This wash protocol was repeated between each protocol step. ELISA plates were then blocked followed by addition of experimental supernatants or standards. Plates were then treated with biotinylated detection antibodies followed by avidin-horseradish peroxidase. 3,3 ′,5,5′-tetramethylbenzidine substrate (eBioscience) was added to each well, with the reaction developed for 10–15 min in the dark. ELISA plates were read at a wavelength of 450 nm with cytokine concentrations calculated using the standard curve values.

Total RNA was isolated using commercial RNAeasy kits (Bioline) according to the manufacturer’s instructions. mDCs were initially stimulated for 4, 8, or 12 h, supernatants were removed, and cells lysed in lysis buffer. Briefly, lysates were homogenized, placed into spin columns, washed, and centrifuged with commercial diluents and RNA was eluted from the spin columns. qRT-PCR was carried out using TaqMan Gene Expression Assays (Applied Biosystems) for IL-1β, IL-6, IL-23, TSLP, and HPRT. RNA was added to master mix consisting of commercial gene specific probe and primer sets (Applied Biosystems), TaqMan^® Universal PCR Master Mix, reverse transcriptase and RNase free H₂O. Taqman reactions were run for 40 cycles on a predefined temperature cycling program. Gene expression for cytokines were normalized to expression of the house-keeping gene, HPRT, and calculated as either relative expression (2^−ΔCt) from unstimulated control cells.

Monocyte-derived dendritic cells were lysed in cytoplasmic lysis buffer [10 mM HEPES, 50 mM NaCl, 0.5 M Sucrose, 0.1 mM EDTA, 0.5% v/v Triton X-100, 10 mM Tetrasodium pyrophosphate, 17.5 mM β-glycerophosphate (all from Sigma) and one complete mini protease inhibitor cocktail tablet (Roche)]. Cellular lysates were scraped, and frozen at −20°C overnight before thawing to aid cell lysis. Lysates were centrifuged at 16,000 g at 4°C and the cytoplasmic supernatant was separated from the nuclear pellet. Total protein was quantified by Bradford Assay (Thermo). SDS-PAGE was used to resolve proteins. Equivalent concentrations of protein lysate were mixed with 6× loading buffer (10% w/v SDS, 0.3 M Tris–HCl, 25% v/v β-Mercaptoethanol) and glycerol (all from Sigma) and boiled for 10 min. Samples and molecular weight reference proteins were loaded on to precast gradient (4–20%) acrylamide gels (BioRad) and run for 90 min at constant amps (30 mA). Resolved proteins were detected after transfer to nitrocellulose membranes by immunoblot. Membranes were blocked in 5% milk (w/v) (Marvel) and washed three times with wash buffer composed of 1× TBS supplemented with 0.1% Tween 20 between each protocol step. Membranes were then probed with primary antibodies to pro-IL-1β (R&D, AB-201-AB), HIF-1α (Novus, NB100-449), and β-actin (Abcam) overnight, and then incubated in HRP-conjugated secondary antibodies. Protein was detected with ECL (PerkinElmer) and visualized using GBox (Syngene). Protein expression was analyzed relative to β-actin control.

Flow cytometry was used to determine mDC activation status, dectin-1 expression, and ROS production. For mDC activation status and dectin-1 expression, mDC were resuspended in FACS buffer consisting of PBS supplemented with 0.1% BSA, 2 mM EDTA, and 0.01% Sodium Azide (All Sigma) after stimulation. ROS were detected by incubating mDC with CellRox Green (Thermo) fluorescence dye during stimulations as per manufacturer’s instructions; followed by harvesting into FACS buffer. mDC were then treated with 10% mouse serum (Sigma) for 30 min on ice to block non-specific binding of antibodies to Fc receptors. mDCs were then stained for 30 min with anti-CD11c (Becton Dickinson, Clone B-ly6), anti-CD86 (Biolegend Clone IT2.2), and anti-HLA-DR (Becton Dickinson, Clone 946.6) to examine mDC activation status and anti-dectin-1 (R&D Systems, Clone 259931) to determine dectin-1 downregulation. Cells were then washed, centrifuged at 300 g for 5 min, and re-suspended in FACS buffer for analysis. mDC expression of CD11c, CD86, HLA-DR, dectin-1, and ROS was determined by mean fluorescence intensity and acquired using a flow cytometer (FACSFortessa, Becton Dickinson).

Cumulative data displayed as mean ± SEM. Data were analyzed using the Prism statistical package (GraphPad) with statistical analysis completed using one-way ANOVA with Bonferroni post-tests (ns = not significant, ****p = 0.001,***p = 0.01, **p = 0.05, *p = 0.1). Flow cytometry data were analyzed using FlowJo 10 software (Treestar).