IL‐38 prevents induction of trained immunity by inhibition of mTOR signaling

Abstract Trained immunity is the acquisition of a hyperresponsive phenotype by innate immune cells (such as monocytes and macrophages) after an infection or vaccination, a de facto nonspecific memory dependent on epigenetic and metabolic reprogramming of these cells. We have recently shown that induction of trained immunity is dependent on IL‐1β. Here, we show that recombinant IL‐38, an anti‐inflammatory cytokine of the IL‐1‐family, was able to induce long‐term inhibitory changes and reduce the induction of trained immunity by β‐glucan in vivo in C57BL/6 mice and ex vivo in their bone marrow cells. IL‐38 blocked mTOR signaling and prevented the epigenetic and metabolic changes induced by β‐glucan. In healthy subjects, the IL1F10 associated single nucleotide polymorphism rs58965312 correlated with higher plasma IL‐38 concentrations and reduced induction of trained immunity by β‐glucan ex vivo. These results indicate that IL‐38 induces long‐term anti‐inflammatory changes and also inhibits the induction of trained immunity. Recombinant IL‐38 could therefore potentially be used as a therapeutic intervention for diseases characterized by exacerbated trained immunity.

increased in bone marrow of β-glucan-treated mice and causative for the induction of trained immunity. 7 In in vitro models of trained immunity, primary human monocytes respond to IL-1β with epigenetic changes that augment the production of inflammatory cytokines such as TNFα and IL-6. 12 Furthermore, the A allele of the single nucleotide polymorphism (SNP) rs16944 in the promotor region of the IL1B gene was associated with increased induction of trained immunity in human monocytes. 12 We have previously reported that IL-38 can inhibit IL-1β, IL-6, and KC in a model of gouty arthritis. 18 As IL-38 can block IL-1β, we hypothesize that IL-38 has an inhibitory role for the induction of trained immunity. In this report, we aimed to define the role of IL-38 in β-glucaninduced trained immunity. We show that IL-38 inhibits the induction of trained immunity in murine models by disrupting the mTOR signaling cascade, which is essential for the induction of trained immunity. 5 In healthy humans, we show that the SNP rs58965312 affects IL-38 plasma levels and the trained immunity response in vitro.

Mice
Seven to nine-week-old male C57Bl/6 mice were obtained from Jackson Laboratories (Bar Harbor, ME, USA). The mice were fed sterilized laboratory chow and water ad libitum. The experiments were approved by the Institutional Animal Care and Use Committees of the University of Colorado Denver (Aurora, CO, USA).

Trained immunity model
Mice received 1 μg of bioactive 18 Wilmington, NC, USA). After anesthetization with isoflurane, blood from the orbital plexus was collected in EDTA, and mice were sacrificed by cervical dislocation. Blood was centrifuged at 1000 ×g for 10 min to prepare plasma for cytokine measurements.

In vitro cytokine production
Cytokine production was assessed ex vivo in whole blood, bone marrow cells, and splenocytes.

Cytokine measurement by ELISA and lactate measurement
Supernatants and plasma were stored at −20 • C until further analysis. ELISA was performed according to the manufacturer's instructions (BioTechne), except the IL-38 ELISA on human plasma was performed as described elsewhere. 20 Mouse plasma and bone marrow were obtained as described and lactate levels were determined by a commercially available Lactate Fluorometric Assay Kit (Biovision, Milpitas, CA, USA).

Western blot
Bone marrow from naïve mice was collected as described earlier.

Chromatin immunoprecipitation
Bone marrow was isolated as described earlier. Cells were fixed in 1% methanol-free formaldehyde and stored at 4 • C. Afterward, cells were sonicated, and immunoprecipitation was performed using antibodies against H3K4me3 (Diagenode, Seraing, Belgium), as described elsewhere. 6 Table S1.

PBMC experiments
The study was performed in a cohort of ±200 healthy individuals of Western European ancestry from the Human Functional Genomics Project (200FG cohorts, see www.humanfunctionalgenomics.org).
PBMCs were isolated from consenting healthy donors as described before. 21 The preparation of plasma for IL-38 ELISA and in vitro stimulations of primary monocyte stimulations in this cohort were described elsewhere. 22 Monocytes were trained with β-glucan for 24 h, washed with warm PBS, and rested in RPMI for 6 d. 23 Then, cells were stimulated with 10 ng/ml LPS. After 24 h, supernatants were collected and stored at −20 • C until IL-6 and TNFα were measured by ELISA. Plasma IL-38 levels were determined by ELISA. The status of IL-38 SNP rs58965312 was assessed as described in the methods of Li et al. 22 by the commercially available SNP chip, Illumina HumanOmniExpressExome-8 v1.0.

Statistical analysis
The differences between the various conditions were analyzed with the Wilcoxon matched pairs signed rank test or Mann-Whitney U-test as appropriate. Data are presented as mean ± SEM unless otherwise indicated. Data were analyzed using GraphPad Prism 8.0 (GraphPad Software, La Jolla, CA, USA). A P-value below 0.05 was considered significant.

IL-38 inhibits LPS-induced inflammation
These studies on the putative role for IL-38 on trained immunity began with an assessment of the effect of recombinant human IL-38 on systemic inflammation due to endotoxemia. Mice received either vehicle or 1 μg of IL-38 i.p. on days 4, 3, and 2 ( Fig. 1A). After 4 d, LPS (5 mg/kg) was administered i.p. to all mice, and 4 h later the mice were sacrificed. As shown in Figure 1B, mice receiving IL-38 exhibited less hypothermia compared to vehicle treated mice, indicating reduced inflammation. 24 Plasma IL-1β, TNFα and IL-6 concentrations were significantly decreased in mice treated with recombinant IL-38 compared to vehicle-treated mice (Fig. 1C), whereas IL-1α concentration in lysates of whole blood was not affected by IL-38. In contrast, no major IL-38-dependent changes in cell counts in bone marrow, splenocytes, and whole blood were observed, apart from a small decrease in lymphocytes in whole blood (data not shown). Whole blood, bone marrow, and splenocytes were further incubated in vitro for 24 h without any additional stimulus. We observed reduced spontaneous release of TNFα (75% decrease) and IL-6 (25% decrease) from whole blood cultures of mice treated with IL-38 (Fig. 1D top). In cultured bone marrow, a reduced spontaneous release of TNFα (55% decrease) was also observed (Fig. 1D middle). Splenocytes did not show differences in IL-6 or TNFα production (Fig. 1D bottom).

IL-38 inhibits β-glucan-induced trained immunity
Next, we assessed whether IL-38 was also able to inhibit the induction of trained immunity by β-glucan. β-glucan-induced trained immunity results in amplified cytokine production to a secondary and unrelated challenge. 7,9 As depicted in Figure 2A β-glucan-trained mice an increase in TNFα and IL-6 production was found, which was abrogated by the addition of IL-38 (Fig. 2D). This cytokine profile was reflected in splenocyte cultures and-although not statistically significantly-in bone marrow (Fig. 2D). Cytokine levels in plasma and blood lysate at sacrifice. (D) Spontaneous TNFα and IL-6 production in supernatants from whole blood, bone marrow, and splenocytes after 24 h ex vivo culture. N = 5, mean ± SEM, cytokine production normalized to monocyte numbers *P < 0.05, **P < 0.01, ns not significant

IL-38 prevents epigenetic histone changes induced by β-glucan
As the heightened induction of inflammatory cytokines by β-glucan is the result of epigenetic changes at the level of histones, 4,9 we next assessed the influence of IL-38 on epigenetic modifications at promoters of established trained immunity-related genes. Mice were treated with β-glucan ± IL-38 and sacrificed after 5 d without LPS challenge (Fig. 3A). 2,6 When whole blood, bone marrow, and splenocytes were stimulated ex vivo with LPS with or without the inflammasome activator nigericin, an increase in cytokine production was seen in β-glucan trained mice, which was abolished by IL-38 (Fig. 3B). When expression of genes related to trained immunity (Tnfa, Nrlp3, Hk2, and Pfkp) was determined in bone marrow of LPS-challenged mice, increased expression of all genes was observed which was prevented by IL-38 (Fig. 3C). To determine whether this is the result of epigenetic modifications induced by β-glucan and IL-38, bone marrow of non-LPS treated mice was processed for epigenetic analysis. H3K4me3 (an important histone marker in trained immunity 4,9 ) marked DNA was isolated, and the percentage of promoters of the trained immunity-related genes that were positive for this histone mark was determined. 2 As shown earlier, an increase in promotors positive for H3K4me3 was seen in βglucan-trained mice, and this was reversed when mice also received IL-38 (Fig. 3D).

IL-38 obstructs β-glucan dependent Akt/mTOR/S6K signaling
We have shown previously that trained immunity in monocytes depends on activation of the mTOR pathway. 5 As IL-38 inhibits the induction of trained immunity, we hypothesized that inhibition of the mTOR pathway takes place, as was also the mechanism for IL-37 inhibition of trained immunity. 17 As shown in Figure 4A

SNP and IL-38 plasma levels predict the strength of trained immunity induction
To validate the influence of IL-38 on trained immunity in humans, we investigated the effect of 11 SNPs associated with IL-38 (IL1F10) and the adjacent IL-1Ra (IL-1RN) on the induction of trained immunity in human PBMC from 118 healthy subjects of the 200FG cohort. 25 We measured plasma IL-38 concentrations, and we quantified the capacity of their monocytes to acquire trained immunity. As shown in Figure 5A, the induction of IL-6 and TNFα after in vitro training with β-glucan is related to the SNP rs5896312. In addition, the plasma level of IL-38 was dependent on this SNP, as carriers of two alternative alleles had higher levels in comparison to subjects carrying two reference alleles (Fig. 5B). No difference was found in plasma IL-1RA levels of these individuals (data not shown). In summary, the inhibition of trained immunity in healthy volunteers was associated with a higher concentration of plasma IL-38, which in turn is dependent on the SNP rs5896312.

DISCUSSION
In this report we demonstrate that IL-38 significantly reduces the induction of trained immunity by β-glucan. This observation is consistent with the anti-inflammatory properties of IL-38 as an inhibitor of innate and adaptive immunity. 16 Percentage of H3K4me3 at the promoters of genes displayed. N = 12, mean ± SEM, cytokine production normalized to monocyte numbers, *P < 0.05, **P < 0.01, ns not significant system, resulting in enhanced proinflammatory responses of the innate immune system upon a second inflammatory challenge. Enhanced production of proinflammatory cytokines also results in epigenetic modifications in the bone marrow, 7 and reduces mortality in models of systemic C. albicans or S. aureus infections. 8,9 Moreover, IL-1β has been shown to play a major role in the induction of trained immunity, 7,12 which is also highlighted by Van der Meer et al., who demonstrated that treating mice with small doses of recombinant human IL-1β reduces mortality in granulocytopenic mice from subsequent Pseudomonas infection, a concept at that time termed nonspecific resistance to infection. 27 Another member of the IL-1 family, IL-37, was shown to inhibit the NLRP3 inflammasome and the release of active IL-1β, and the induction of trained immunity. 17 Given the anti-inflammatory properties of IL-38, 28 such as its ability to block IL-1β upregulation in mouse models of gouty arthritis, 18 we investigated whether IL-38 can block the induction of trained immunity.
Here, we show that in addition to preventing LPS-induced inflammation in mice, 29  IL-38 binds to the IL-36R, and IL-1R9 is its putative coreceptor. 16,30 It has been reported that N-terminally truncated IL-38 attenuates the JNK/AP1 pathway which results in reduced IL-6 production from macrophages. 30 Here we show that IL-38 also inhibits the Akt/mTOR pathway, which is essential in the induction of trained immunity. 5 Interestingly, inhibition of the mTOR pathway has also been observed for IL-37, 31,32 which suggests a partially comparable working mechanism of these broad anti-inflammatory IL-1 family members.
Besides the positive role of trained immunity in providing protection against infection, a deleterious effect has been suggested in auto-immune diseases. 3 Therefore, IL-38 may be used therapeutically to temper trained immunity in these diseases. Trained immunity was specifically suggested to play a pathologic role in systemic lupus erythematosus (SLE), Sjögren's syndrome and rheumatoid arthritis. 3 A protective role of IL-38 was proposed in SLE, 26 in mice subjected to experimental arthritis, 33 and in Sjögren's disease IL-38 was upregulated as a counterbalance to IL-36-dependent inflammation. 34 Hence, a novel hypothesis is that recombinant IL-38 could be used to inhibit inflammation in auto-immune diseases by inhibiting trained immunity.
Trained immunity was also suggested to play a role in auto- This study presents evidence that IL-38 inhibits the induction of trained immunity by interfering with the mTOR pathway and with IL-1 signaling. Apart from interfering with these essential signaling pathways in the induction of trained immunity by β-glucan, IL-38 itself also induced epigenetic changes, which suggests that separate from inhibiting trained immunity IL-38 might have a long-term immunosuppressive effect. The role of IL-38 in human diseases that depend on trained immunity requires further characterization, for example by assessment of human samples to determine the effect of IL-38 on inflammation in these diseases' contexts. The metabolic effects of IL-38 and its effects on trained immunity become especially relevant in the setting of cardiovascular disease. Trained immunity is thought to contribute to persistent inflammation that characterizes atherosclerosis. 36 One marker of this persistent inflammation is C-reactive protein (CRP) and in a meta-analysis of more than 80,000 subjects there were polymorphisms found in only five immune genes that associated with circulating CRP, which included CRP itself, NRLP3, IL-6R, and strikingly IL-1F10 (IL-38). 37 We reported recently that IL-38 plasma concentrations correlate inversely with IL-6 and CRP in overweight subjects, and are most reduced in subjects with metabolic syndrome. 20 These and our current observations suggest that IL-38 might be a key player in restraining metabolic pathways and inflammation that are crucial drivers of cardiovascular disease.
In conclusion, in this study we show that IL-38 is an important anti- contributed reagents/materials. D.M.G. and R.J.W.A. wrote/drafted and finalized the paper. All authors read and approved the final manuscript.