TNF inhibitors have been used to treat autoimmune and autoinflammatory diseases. Here we report an unexpected mechanism underlying the therapeutic effects of TNF inhibitors in Behçet’s

disease (BD), an autoimmune inflammatory disorder. Using serum metabolomics and peripheral immunocyte transcriptomics, we find that polymorphonuclear neutrophil (PMN) from patients with BD

(BD-PMN) has dysregulated mevalonate pathway and subsequently increased farnesyl pyrophosphate (FPP) levels. Mechanistically, FPP induces TRPM2-calcium signaling for neutrophil extracellular

trap (NET) and proinflammatory cytokine productions, leading to vascular endothelial inflammation and damage. TNF, but not IL-1β, IL-6, IL-18, or IFN-γ, upregulates TRPM2 expression on

novel insights for TNF inhibitor therapies on BD.

Different from the well-characterized damage-associated molecular pattern (DAMP), metabolic intermediates, have recently been identified as a novel class of endogenous danger signals that

strongly elicit a variety of biological functions1,2,3,4. In response to external stimulations, polymorphonuclear neutrophil (PMN) tend to adopt specific metabolic pathways5, for a tuned

purpose to support specialized effector functions such as neutrophil extracellular trap (NET) formation (NETosis)6,7. Hence, considering the short cell lifespan8, PMN is a potentially

important source of serum danger signals during the shift of its metabolic patterns9. However, the immunometabolism of PMN in autoimmune and autoinflammatory diseases remains to be

investigated.

Corresponding to the clinical features of autoinflammatory and autoimmune diseases, Behçet’s disease (BD) is a chronic systemic vasculitis characterized by recurrent oral or genital ulcers,

skin lesions, and involvement of vital organs, such as ocular, cardiovascular, gastrointestinal, and neurological manifestations10. BD is sight-threatening and even life-threatening,

imposing considerable financial burdens on society and individuals. BD is a representative type of immune disorder as it exhibits aberrant and excessive activation of both innate and

adaptive immunity, which is thus considered as a unique and crucial clinical condition linking both autoimmunity and autoinflammation11. Although the etiology of BD is still unknown, it is

well accepted that during the progression of BD, an interplay between endogenous danger signals and PMN is of note12,13,14. In this regard, elevated serum DAMPs such as high mobility group

box 1 (HMGB1)15 and S100 calcium-binding protein A12 (S100A12)16 have been reported in BD. These DAMPs promote acute inflammation and recruitment of PMN to vascular lesions17, and thus BD

was also recognized as a “neutrophilic vasculitis”18. Moreover, PMN is particularly prone to undergo cell necrosis, mainly in the form of NETosis under inflammatory conditions19,20, which

involves the release of various DAMPs, including DNA, histones19, HMGB1, and S100A8, etc., aggravating the activation and inflammation of macrophages19 and vascular endothelium20,21 in BD.

As an integrating mechanism, all these accounts for the production of proinflammatory cytokines, including tumor necrosis factor (TNF), interferon-γ (IFN-γ), interleukin 6 (IL-6), and skewed

T-helper (Th) 1 and Th17 cell activation22. Thus, BD is ideal for use as a representative disease to investigate the mechanism of action underlying the therapeutic effects of TNF inhibitors

in treating autoimmune and autoinflammatory diseases. Farnesyl pyrophosphate (FPP), a key metabolite in the mevalonate (MVA) pathway, plays a crucial role in cholesterol biosynthesis, has

been implicated in inflammatory responses, and potentially contributed to autoinflammatory and autoimmune diseases23,24. However, its implications in BD pathogenesis remain largely to be

elucidated.

Here in this study, we conduct comprehensive analyses of serum metabolomics and peripheral immune cell transcriptomics, revealing a dysregulated mevalonate pathway in PMN from BD patients

(BD-PMN) and subsequently elevated FPP levels. Further investigations demonstrated that FPP promotes BD-PMN hyperactivation via a calcium-TRPM2-dependent pathway. TNF upregulates TRPM2

expression on BD-PMN, while TNF inhibitors have the opposite effect. Our findings highlight the potential pathogenic involvements of FPP in BD and uncover the immunometabolic mechanisms

underlying disease progression. These insights provide novel therapeutic implications for TNF inhibitors in BD and potentially other autoimmune and autoinflammatory disorders.

To gain an unbiased understanding of immunometabolic profiles in BD, we first integrated multi-omic analyses in our previously published cohorts containing BD patients and the sex- and

age-matched HC, including serum metabolomics and lipidomics25, bulk RNA-sequencing of peripheral blood mononuclear cell (PBMC) (GSE19853311), PMN (GSE20586712), and single-cell

RNA-sequencing of PBMC (GSE19861611). We found that the steroid biosynthesis pathway is significantly upregulated in BD serum in comparison to that from HC (Fig. 1A, Supplementary Fig. 1A).

Multi-omic analyses of individual immune cell populations revealed that this metabolic alteration in serum was mainly displayed in the BD-PMN (Fig. 1B, and Supplementary Fig. 1B).

Considering that there are multiple downstream pathways in steroid synthesis, we then conducted an enrichment analysis of each pathway. We found that the cholesterol biosynthesis pathway,

but not the synthesis of bile acid and steroid hormone pathways, was significantly increased in BD-PMN in comparison to HC-PMN (Fig. 1C). It shall be noted that although our single-cell

sequencing analyses also indicated a slight upregulation of the steroid synthesis pathway in monocytes from BD patients, their cholesterol synthesis pathway was not significantly upregulated

as in the case of PMN (Supplementary Figs. 1C, D).

A Lipidomic analysis between treatment-naïve BD and HC serum. B GSEA of steroid biosynthesis pathway in BD and HC peripheral blood immune cells, including bulk RNA-seq of PBMC (GSE198533),

PMN (GSE205867), and single-cell RNA-seq of PBMC (GSE198616). C GSEA of steroid synthesis sub-pathways, including bile acids, steroid hormones, vitamin D, and cholesterol synthesis, in BD

and HC PMN (GSE205867). D Heatmap showing enzyme expression of cholesterol synthesis pathway in BD PMN RNA-seq (GSE205867). Significantly increased and decreased genes are marked in red and

blue, respectively. E Schematic plot showing the intervention of the cholesterol synthesis pathway. Significantly elevated enzymes validated by both BD PMN RNA-seq and qRT-PCR (N = 20) are

marked in red. The agonists and inhibitors are marked by green and purple arrows, respectively. N for all experiments were biological replicates. ZGA, zaragozic acid; SQS, squalene synthase.

Next, we examined the expression profiles of crucial enzymes in the cholesterol biosynthesis pathway from transcriptional datasets of BD-PMN and HC-PMN (Fig. 1D), and further validated the

transcription levels of these target genes by qRT-PCR in an independent cohort of twenty BD patients (Supplementary Fig. 1E, and Supplementary Table 1). We found that those most

differentially expressed enzymes, including ACAT1, MVK, PMVK, and MVD, were concentrated in the upstream pathway of cholesterol biosynthesis, i.e., the MVA pathway (Fig. 1E). These results

reveal an upregulated MVA pathway specifically in BD-PMN.

To determine whether the upregulated MVA pathway is involved in the hyperactivation of BD-PMN, we inhibited 3-hydroxy-3-methylglutaryl-CoA reductase (HMGCR) and farnesyl pyrophosphate

synthase (FPPS, also known as farnesyl diphosphate synthase (FDPS)), which control the start and end points of the MVA pathway, via the clinically-available drugs simvastatin and zoledronic

acid, respectively (Fig. 1E). Notably, both inhibitors suppressed proinflammatory cytokines in BD-PMN (Fig. 2A, B). Since FPP is the end product of the MVA pathway that is converted by FPPS

enzyme from geranyl-pyrophosphate (GPP), we speculated that FPP might be the primary component participating in the MVA pathway-induced BD-PMN hyperactivation. To confirm this hypothesis, we

either inhibited FPP-metabolizing enzymes farnesyl-diphosphate farnesyltransferase (FDFT1) (squalene synthase), via zaragozic acid (ZGA) and BPH-652, or activated FDFT1 function via

ferroptosis inducing 56 (FIN56)26. The results demonstrated that intracellular accumulation of FPP increased the production of proinflammatory cytokines in BD-PMN (Fig. 2C, D), and vice

versa (Fig. 2E), both of which suggested a proinflammatory role of FPP in PMN. Notably, none of the aforementioned inhibitors or agonists had any effect on PMN viability (Supplementary Figs.

 2A, B). To conclude, these results suggested a proinflammatory role of the MVA pathway, especially its metabolite FPP, in PMN activation and inflammation.

A–E BD-PMN was pretreated with either simvastatin (1 μM and 10 μM, A) to inhibit HMGCR, zoledronic acid (1 μM and 10 μM, B) to inhibit FDPS, ZGA (7.5 μM and 15 μM, (C) and BPH652 (40 μM and

80 μΜ, (D) to inhibit squalene synthase FDFT1, or FIN56 (1 μM and 5 μM, E) to activate FDFT1, for 16 hours. The expression levels of TNF, IL-6, IL-18, and IL-1β were detected by qRT-PCR (N =

 4, for each group). F Relative abundance of FPP in PMN (N = 24) and serum (N = 23) from active BD and HC. G, H Relative abundance of FPP in PMN (N = 15, G) and serum (N = 14, H) from active

and remission BD. I, J Correlations between the relative abundance of FPP in BD-PMN (N = 39, (I) and serum (N = 37, J) with BD clinical activity indicators, CRP and ESR. K, L Relative

abundance of FPP in PMN (N = 32 vs 7, (K) and serum (N = 23 vs 14, L) of BD with different BDCAF. N for all experiments were biological replicates. Data are presented as mean ± SD; error

bars indicate the SD. A two-sided p-value