A TNFSF15 disease-risk polymorphism increases pattern-recognition receptor-induced signaling through caspase-8–induced IL-1

Significance

This study describes a role for tumor necrosis factor ligand superfamily member 15 (TNFSF15), a molecule being considered for therapy in inflammatory diseases. We identify that death receptor 3 (DR3) is clearly expressed on human macrophages and define a new role for TNFSF15:DR3 signaling in mediating innate immune outcomes in human myeloid-derived cells, as well as mechanisms mediating these outcomes. We further define the disease-relevant, functional consequences of polymorphisms in the TNFSF15 region, one of the 163 inflammatory bowel disease risk loci. To our knowledge, our studies highlight a previously unknown mechanism through which TNFSF15:DR3 can contribute to intestinal inflammation, namely by enhancing signaling and cytokine secretion upon stimulation of a broad range of pattern-recognition-receptors, mycobacterial components, and live bacteria.

Abstract

Inflammatory diseases are characterized by dysregulated cytokine production. Altered functions for most risk loci, including the inflammatory bowel disease and leprosy-associated tumor necrosis factor ligand superfamily member 15 (TNFSF15) region, are unclear. Regulation of pattern-recognition-receptor (PRR)-induced signaling and cytokines is crucial for immune homeostasis; TNFSF15:death receptor 3 (DR3) contributions to PRR responses have not been described. We found that human macrophages expressed DR3 and that TNFSF15:DR3 interactions were critical for amplifying PRR-initiated MAPK/NF-κB/PI3K signaling and cytokine secretion in macrophages. Mechanisms mediating TNFSF15:DR3 contributions to PRR outcomes included TACE-induced TNFSF15 cleavage to soluble TNFSF15; soluble TNFSF15 then led to TRADD/FADD/MALT-1– and caspase-8–mediated autocrine IL-1 secretion. Notably, TNFSF15 treatment also induced cytokine secretion through a caspase-8–dependent pathway in intestinal myeloid cells. Importantly, rs6478108 A disease risk-carrier macrophages demonstrated increased TNFSF15 expression and PRR-induced signaling and cytokines. Taken together, TNFSF15:DR3 interactions amplify PRR-induced signaling and cytokines, and the rs6478108 TNFSF15 disease-risk polymorphism results in a gain of function.

Immune-mediated diseases, such as inflammatory bowel disease (IBD), exhibit dysregulated cytokines (1). In the intestine, regulating pattern-recognition receptor (PRR) signaling and cytokine secretion is crucial given ongoing host:microbial interactions; dysregulated PRR-initiated signaling can exacerbate mouse colitis, and contribute to human IBD (2–4). Multiple genetic loci regulate immune-mediated diseases. However, functional outcomes for most disease-risk loci are unclear, yet crucial to the understanding necessary for disease mechanisms and therapy.

Polymorphisms in a region that includes TNFSF15 are associated with various diseases (5, 6), including IBD (3, 7, 8) and leprosy (9). TNFSF15 neutralization ameliorates DSS-induced mouse colitis (10). This amelioration has been attributed to prevention of myeloid cell-derived TNFSF15 interactions with death receptor 3 (DR3)-expressing T cells (11, 12); decreased signaling from DR3 then regulates T-cell activation/differentiation (10). TNFSF15 is up-regulated on macrophages from IBD patients (11) and is being considered for therapeutic targeting in IBD. DR3-mediated outcomes in myeloid cells, and more specifically, TNFSF15:DR3 regulation of PRR-initiated signaling and cytokine secretion in myeloid cells, have not been described. However, as PRR mediate host:microbial interactions, and myeloid cells are critical in intestinal inflammation (2), we hypothesized that TNFSF15:DR3 might regulate PRR-initiated signaling and cytokines from myeloid cells and that disease-associated TNFSF15 polymorphisms would affect TNFSF15-mediated myeloid cell outcomes.

We found that TNFSF15:DR3 interactions dramatically amplified PRR- and mycobacterial antigen-induced cytokines in human monocyte-derived macrophages (MDM) and monocyte-derived dendritic cells (MDDC) and bacterial-induced cytokines in intestinal myeloid cells ex vivo. Signaling downstream of DR3, rather than TNFSF15, was required for nucleotide-binding oligomerization domain 2 (NOD2)-induced MAPK, NF-κB, and PI3K activation and cytokine secretion. Soluble TNFSF15, produced by TNF converting enzyme (TACE)-mediated cleavage of transmembrane TNFSF15, was sufficient for cytokine amplification in MDM. TNFSF15:DR3 interactions amplified additional cytokines through TNF receptor type 1-associated Death domain protein (TRADD)/Fas-Associated protein with Death Domain (FADD)/mucosa-associated lymphoid tissue lymphoma translocation protein 1 (MALT1) and caspase-8–dependent, but caspase-1–independent, early IL-1 secretion. Finally, MDM from rs6478108 A risk carriers in the TNFSF15 region showed increased TNFSF15 expression, and increased NOD2-induced signaling and cytokines relative to GG carriers. Thus, TNFSF15:DR3 interactions amplify PRR-initiated outcomes, and TNFSF15 risk polymorphisms are gain of function, thereby highlighting therapeutic potential for targeting TNFSF15:DR3.

Results

The TNFSF15 Region Disease-Risk Polymorphism Increases PRR-Induced Cytokines in Primary Human Myeloid Cells.

Myeloid-derived cells and PRR-initiated outcomes are important contributors to IBD pathophysiology (1). We therefore asked whether immune disease-associated TNFSF15 region polymorphisms modulate PRR-induced cytokines in human MDM. Given the NOD2 associations to Crohn disease (2), we treated MDM from 100 healthy individuals with muramyl dipeptide (MDP), the minimal bacterial peptidoglycan component activating NOD2 (13, 14). We examined IL-1β secretion, which strongly amplifies PRR-mediated signaling and cytokine secretion in MDM (15) and is elevated in tissues from IBD patients (16). We normalized IL-1β secretion to untreated cells and log₂ transformed the data. We chose four IBD- (3, 7, 8) or leprosy-associated (9) TNFSF15 region polymorphisms with reasonable frequencies in the healthy population. Of these polymorphisms, the rs6478108 polymorphism most significantly modulated NOD2-induced IL-1 secretion (Fig. S1A). MDM from rs6478108 A risk carriers secreted more IL-1β compared with GG carriers (Fig. 1A). This increase was most pronounced at low MDP doses (Fig. 1A). The antiinflammatory cytokine IL-10 was regulated similarly (Fig. S1B).

Fig. 1.

Human myeloid cells from disease-associated rs6478108 A carriers demonstrate increased PRR-induced cytokine secretion. MDM (n = 100) were treated for 24 h with 1, 10, or 100 μg/mL MDP (A), or 1, 10, or 100 μg/mL Pam₃Cys (B). MDDC (n = 98) were treated for 24 h with 1 μg/mL MDP (NOD2), 1 μg/mL Pam₃Cys (TLR2), 0.1 μg/mL polyI:C (TLR3), 0.01 μg/mL lipid A (TLR4), 0.5 ng/mL flagellin (TLR5), 0.1 μg/mL CL097 (TLR7), or 0.1 μg/mL CpG DNA (TLR9) for 24 h alone (C) or in combination (D). Fold IL-1β induction (log₂ transformed) upon PRR stimulation + SEM stratified on rs6478108 genotype. *P < 0.05; **P < 0.01; ***P < 0.001.

Gram-positive bacteria contain both peptidoglycan (stimulates NOD2) and lipotechoic acid [stimulates Toll-like receptor (TLR) 2]. Rs6478108 A carrier MDM secreted increased cytokines following TLR2 stimulation (Fig. 1B and Fig. S1C). The polymorphism affected NOD2- and TLR2-stimulated MDM (Fig. 1 A and B and Fig. S1 B and C) and MDDC (Fig. 1C and Fig. S1D) similarly. Microbial products activate multiple PRR, and NOD2 can synergize with other PRR (17). In a separate cohort of 98 healthy individuals, rs6478108 A risk carrier MDDC secreted increased IL-1β and IL-10 upon stimulation of multiple TLR alone or combined with NOD2 (Fig. 1 C and D and Fig. S1 D and E). Taken together, risk polymorphisms in the TNFSF15 region, a locus associated with multiple immune-mediated diseases, increase PRR-induced cytokines in myeloid-derived cells.

TNFSF15 and DR3 Are Expressed on MDM.

Because TNFSF15 region polymorphisms regulate PRR-induced cytokines in isolated myeloid cells, we hypothesized that myeloid cells express both TNFSF15 and its binding partner, DR3. Studies to date have not examined DR3 in myeloid cells. We observed low but significant DR3 expression on MDM (Fig. 2A), which was unchanged following MDP treatment. We confirmed antibody specificity by verifying DR3 expression in activated T cells (Fig. 2B) (11) and decreased surface DR3 following siRNA to DR3 on MDM (Fig. 2C). A second DR3 antibody showed similar results. TNFSF15 is expressed on human MDM (18). Surface TNFSF15 increased following NOD2 stimulation, peaking 12–24 h after treatment with 100 μg/mL MDP (dosed as per prior studies; refs. 15, 17, 19, and 20) (Fig. S2A). We verified antibody specificity by TNFSF15 attenuation upon TNFSF15 knockdown (Fig. S2B). TNFSF15 induction occurred transcriptionally, with the canonical TNFSF15 147-bp isoform peaking 4 h after MDP treatment (Fig. S2C). Surface TNFSF15 can be processed to soluble TNFSF15; NOD2 stimulation increased soluble TNFSF15 (Fig. S2D). Therefore, DR3 is expressed on MDM, and NOD2 stimulation increases surface and soluble TNFSF15 protein.

Fig. 2.

Human MDM express DR3. (A, Left) Representative flow cytometry with DR3 (solid line) MFI values in MDM. (Right): Summary graph (n = 8). (B) PBMC were left untreated (dotted line) or treated with 50 ng/mL PMA and 200 ng/mL ionomycin (solid line) for 24 h and gated on CD3⁺ T cells. (C) MDM (n = 4) were transfected with scrambled (solid line) or DR3 (dotted line) siRNA. (C, Left) Representative flow cytometry with DR3 MFI values. (C, Right): Summary graph (n = 4). Isotype controls (gray shading). iso, isotype. ††, P < 1 × 10⁻⁵.

TNFSF15 and DR3 Amplify Cytokines by PRR Ligands and Mycobacterial Components.

We next sought to establish that each TNFSF15 and DR3 is required for regulating PRR-induced cytokines. Upon knockdown of either TNFSF15 (Fig. S2B) or DR3 (Fig. 2C), NOD2-induced proinflammatory and antiinflammatory cytokine secretion was reduced (Fig. 3A). We confirmed these results through additional siRNAs to TNFSF15 and DR3 (Fig. S3A) and using DR3 blockade (Fig. 3B). Cells were viable (Fig. S3B), and antiinflammatory cytokine secretion through dectin ligands was not affected under siRNA (Fig. S3C) or antibody neutralizing conditions (Fig. S3D), indicating pathway selectivity and cell functionality. Moreover, cytokine secretion induced by multiple PRR was regulated by TNFSF15 and DR3 (Fig. S3E). TNFSF15 region polymorphisms are associated with leprosy (9); TNFSF15 (Fig. S4A) and DR3 (Fig. S4B) were required for optimal cytokine secretion by whole cell Myobacterium leprae lysate and its cytosolic, membrane, and total lipid fractions. The dependency on TNFSF15 and DR3 extended to additional mycobacteria, specifically Myobacterium tuberculosis components (Fig. S4). We examined the physiological significance of cytokine amplification by TNFSF15 in human intestinal myeloid cells given the TNFSF15 associations with IBD. Intestinal myeloid cells secrete IL-1 following live Salmonella enterica serovar typhimurium exposure (21, 22), and this autocrine IL-1 in turn contributes to IL-8 secretion (22). DR3 blockade dramatically attenuated S. typhimurium-induced cytokines in intestinal myeloid cells (Fig. 3C) and in MDM (Fig. S3F). Therefore, TNFSF15 and DR3 amplify cytokine secretion by multiple PRR, mycobacterial components, and live bacteria in MDM, and by pathogenic bacteria in intestinal myeloid cells.

Fig. 3.

Stimulation of DR3 with TNFSF15 is required for optimal PRR-induced cytokine secretion and signaling. (A) MDM (n = 4) were transfected with scrambled, TNFSF15, or DR3 siRNA for 48 h, then treated with 100 μg/mL MDP for 24 h. (B) MDM (n = 4) were incubated with anti-DR3 (1 μg/mL) blocking antibody or isotype control (mIgG1) for 1 h, then treated with 100 μg/mL MDP for 24 h. Similar results were observed for an additional n = 4. (C) Intestinal myeloid cells (n = 4) were incubated with blocking anti-DR3 or isotype control as in B, then cultured with 10:1 MOI S. typhimurium (S. typh) for 24 h. (D) MDM (n = 4) were treated for 24 h with 10 ng/mL recombinant TNFSF15 ± 1 μg/mL MDP. (A–D) Shown is cytokine secretion + SEM. Synergy was calculated as fold increase in cytokine levels upon combined TNFSF15 and MDP treatment divided by the sum of the cytokine levels upon treatment with each stimulus alone. (E) MDM (n = 8) were transfected with scrambled or TNFSF15 siRNA. Forty-eight hours later, cells were treated for 15 min with 100 μg/mL MDP. (E, Left) Representative flow with MFI values for phospho-ERK. (E, Right) Fold phospho-ERK induction + SEM. scr, scrambled; T15, TNFSF15; Tx, treatment. Lines over adjacent bars indicate identical P values. *P < 0.05; **P < 0.01; ***P < 0.001; †, P < 1 × 10⁻⁴; ††, P < 1 × 10⁻⁵.

TNFSF15 Stimulates DR3 To Initiate Downstream Signaling Required for Amplifying PRR-Induced Cytokine Secretion.

We next asked whether TNFSF15-induced amplification of PRR-initiated cytokines was mediated via TNFSF15- or DR3-initiated signaling. Stimulation under suboptimal conditions can bring out contributions of cytokine-amplifying stimuli; 1 μg/mL MDP induces low cytokine secretion (Fig. 1A and Fig. S1B), such that we questioned whether TNFSF15 or DR3, respectively, augments this low secretion. Recombinant DR3 failed to enhance MDP-induced cytokines (Fig. S3G), indicating that TNFSF15-initiated signaling was not enhancing NOD2 responses. Recombinant DR3 was functional, as it competed to attenuate TNFSF15-induced cytokine secretion (Fig. S3H). In contrast, soluble TNFSF15 alone induced low-level cytokine secretion and amplified MDP-induced cytokines (Fig. 3D). To clearly establish that TNFSF15 and DR3 are interacting with each other, we confirmed that DR3 knockdown attenuated TNFSF15-induced cytokines (Fig. S3I). Therefore, TNFSF15 stimulates DR3-initiated signaling on MDM which then amplifies NOD2-induced cytokines.

TNFSF15 Induces MAPK, NF-κB, and PI3K Signaling.

We next investigated which signaling pathways leading to cytokine secretion were induced by TNFSF15:DR3 interactions in MDM. NOD2 stimulation activates the MAPK ERK, p38, and JNK (15, 17), and each of these pathways is required for NOD2-induced cytokine secretion by human myeloid cells (22). Whether DR3 stimulation activates MAPK in MDM is unknown. Low-dose MDP treatment and TNFSF15 treatment each weakly activated all three MAPK; combined MDP and TNFSF15 treatment enhanced this activation (Fig. S5A). Each of these kinases was required for the cytokine secretion observed (Fig. S5B). NF-κB and PI3K pathways also mediate PRR-induced cytokines (15, 17), and NOD2 and DR3 synergized in NF-κB (Fig. S5C) and PI3K (Fig. S5D) pathway activation. Thus, NOD2 and DR3 synergize to activate MAPK, NF-κB, and PI3K pathways.

TNFSF15:DR3 Signaling Is Required for Optimal MDP-Initiated MAPK, NF-κB, and PI3K Activation.

Exogenous TNFSF15 synergizes with suboptimal 1 μg/mL MDP treatment to activate signaling pathways (Fig. S5). We therefore questioned whether endogenous TNFSF15 amplified signaling during optimal 100 μg/mL MDP treatment. TNFSF15 knockdown significantly attenuated MAPK, NF-κB, and PI3K pathway activation during these conditions (Fig. 3E and Fig. S6), indicating that TNFSF15 is required for MDP-initiated signaling.

TACE Processes TNFSF15 To Amplify PRR-Induced Cytokines.

We next investigated the requirements for transmembrane versus soluble TNFSF15 in amplifying PRR-induced signaling. TNFSF15 knockdown decreases MDP-induced signaling within the first 15 min (Fig. 3E). This short time frame implicates preexisting TNFSF15 in amplifying MDP-initiated signaling. Consistently, MDM express transmembrane TNFSF15 at baseline (Fig. S2A), and MDP treatment induced detectable early (15 min) soluble TNFSF15 upon preventing consumption with DR3 blockade (Fig. S7A). The transcriptional inhibitor actinomycin did not decrease early TNFSF15 processing (Fig. S7A), confirming processing from preexisting transmembrane TNFSF15. We confirmed actinomycin functionality through its inhibition of PRR-induced cytokine transcripts. TNFSF15 cleavage requirements are controversial and can differ depending on cell type (23, 24). To examine whether TACE cleaves TNFSF15 in MDM, we inhibited TACE by TAPI-1. This inhibition abolished NOD2-induced early (15 min) (Fig. 4A) and late (24 h) (Fig. S7B) soluble TNFSF15, thereby increasing surface TNFSF15 (Fig. S7C). Despite increased surface TNFSF15, TACE inhibition in 24-h MDP-treated MDM decreased secretion of multiple cytokines (Fig. 4B), indicating that surface TNFSF15 is not sufficient to amplify NOD2-initiated cytokines. Such differences between membrane-bound and soluble forms have been observed for other cytokines (25). Because TACE can process multiple surface proteins (26), we confirmed that soluble recombinant TNFSF15 rescued the reduced MDP-induced cytokines during TACE inhibition (Fig. 4B). Therefore, soluble, but not membrane, TNFSF15 is sufficient to amplify NOD2-induced cytokines.

Fig. 4.

Soluble TNFSF15 is necessary for amplification of NOD2-induced cytokines. MDM (n = 4) were incubated with 10 μM TAPI-1 and treated with 100 μg/mL MDP for 15min following anti-DR3 (1 μg/mL) blocking antibody pretreatment (prevents early TNFSF15 consumption) (A) or 24 h with or without 10 ng/mL TNFSF15 (B). Shown are cytokines in supernatants. *P < 0.05; **P < 0.01; ***P < 0.001; †, P < 1 × 10⁻⁴; ††, P < 1 × 10⁻⁵.

We next examined how NOD2 regulates TACE. TACE expression was unchanged 15 min after MDP treatment (Fig. S7D), a time at which soluble TNFSF15 is detected. However, TACE expression decreased by 4 h and was maximally decreased 8 h after NOD2 stimulation (Fig. S7D). This decrease at 8 h depended on endocytosis and V-ATPases (Fig. S7E), which contribute to intraendosomal acidification. We next examined mechanisms for NOD2-induced early TACE activation. The MAPK p38 can induce TACE activity (27). ERK, p38, JNK, and NF-κB activation were required for NOD2-induced early TACE activity and TNFSF15 processing (Fig. S7 F and G). Consistent with TNFSF15 activating MAPK and NF-κB (Fig. S5), TNFSF15 knockdown decreased TACE activity (Fig. S7H). Therefore, autocrine TNFSF15 can modulate its own cleavage through regulating TACE activity. Thus, TACE-mediated generation of soluble TNFSF15 amplifies PRR-induced cytokines, and NOD2-induced TACE activity depends on MAPK, NF-κB, and TNFSF15 itself.

TIMP-3 Inhibits NOD2-Induced TNFSF15 Processing.

Tissue inhibitor of metalloproteinases (TIMP)-3 overexpression in mouse myeloma cells inhibits TACE (28). To determine whether endogenous TIMPs regulate TACE-mediated TNFSF15 processing in PRR-stimulated MDM, we knocked down TIMP-1, TIMP-2, and TIMP-3 (Fig. S8A). TIMP-3, but not TIMP-1 or TIMP-2, knockdown increased early and long-term soluble TNFSF15 (Fig. S8B) and decreased surface TNFSF15 (Fig. S8C), indicating that TIMP-3 inhibits TACE-mediated TNFSF15 cleavage. Consistently, TIMP-3 knockdown increased early TACE activity (Fig. S8D). NOD2-induced early TACE activity and TNFSF15 processing were not associated with altered TIMP-3 surface or cellular protein expression (Fig. S8 E and F). However, TIMP-3 protein and mRNA expression increased at later times (Fig. S8 E and G), consistent with cytokine secretion cessation at later time points after PRR stimulation (19). PRR-induced IL-10 was implicated in TIMP3 induction in human monocytes (29). We found that IL-10 and TGF-β each contributed to optimal TIMP3 protein and mRNA expression 24 h after NOD2 stimulation (Fig. S8 H and I). Therefore, TIMP-3 inhibits TACE-mediated TNFSF15 processing in MDM.

TNFSF15:DR3 Signals Through TRADD/FADD/MALT-1/Caspase-8 To Process Pro–IL-1 and Amplify NOD2-Initiated Cytokines.

Autocrine cytokine loops can cooperate to amplify PRR-initiated signaling. Besides TNFSF15, rapid pro–IL-1 processing to secreted IL-1 amplifies PRR-induced cytokines (15). Because long-term (24 h) DR3 stimulation induced IL-1 secretion (Fig. 3D), we hypothesized that early IL-1 secretion might be critical for the TNFSF15:DR3-mediated amplification of other cytokines. Upon preventing early IL-1 consumption through IL-1Ra, we detected IL-1 secretion within 15 min of DR3 stimulation (Fig. 5A). This secreted IL-1 was the mature IL-1 isoform (Fig. S9A); mature IL-1 was similarly detected in cell lysates (Fig. S9A). Secreted IL-1 was processed from existing pro–IL-1 in MDM (Fig. S9B); consistently transcriptional blockade did not affect early DR3-mediated IL-1 secretion (Fig. 5A). Importantly, DR3-induced IL-1 amplified secretion of additional cytokines (Fig. 5B).

Fig. 5.

TNFSF15:DR3 signaling through TRADD/FADD/MALT1/caspase-8 induces early IL-1 secretion that amplifies additional cytokines. (A) MDM (n = 4) were treated with 10 ng/mL TNFSF15 for 15 min. Conditions include IL-1Ra (0.5 μg/mL) (prevents early IL-1 consumption) or actinomycin D (10 μg/mL) (prevents transcription) pretreatment for 1 h. Supernatants were assessed for early IL-1. (B) MDM (n = 4) were pretreated with 0.5 μg/mL IL-1Ra and then with 10 ng/mL TNFSF15 for 24 h. Shown are cytokine concentrations. (C) MDM were treated with 10 ng/mL TNFSF15 or 100 μg/mL MDP for 15 min. Active caspase-8 and caspase-1 expression by Western blot was from two of three donors. GAPDH was assessed on separate blots. (D–G) MDM (n = 4) were transfected with scrambled, caspase-1, caspase-8, caspase-3, or caspase-4 siRNA, or combined caspase-1 and caspase-8 siRNA. Forty-eight hours later, cells were treated with 10 ng/mL TNFSF15 (D and F) or 100 μg/mL MDP (E and G) for 15 min following 0.5 μg/mL IL-1Ra pretreatment (D and E) or 24 h (F and G). Shown are cytokine concentrations. (H) Intestinal myeloid cells (n = 4) were pretreated with 50 μM YVAD (caspase-1 inhibitor) or 50 μM caspase-8 inhibitor II and then treated with 10 ng/mL TNFSF15 or 100 μg/mL MDP for 24 h. Shown are cytokine concentrations. (I) DR3 signaling model. (J and K) MDM (n = 4) were transfected with scrambled, TRADD, FADD or MALT1 (Left), or TRAF2, RIP1 or RIP3 (Right) siRNA. Forty-eight hours later, cells were treated with IL-1Ra (0.5 μg/mL) (prevents early IL-1 consumption) and then treated with 10 ng/mL TNFSF15 for 15 min (J) or 10 ng/mL TNFSF15 for 24 h (K). Shown are cytokine concentrations. Results were confirmed in another n = 4. scr, scrambled; T15, TNFSF15; Tx, treatment. *P < 0.05; **P < 0.01; ***P < 0.001; †, P < 1 × 10⁻⁴; ††, P < 1 × 10⁻⁵.

We next investigated mechanisms required for early TNFSF15:DR3-induced IL-1 processing. TNFSF15:DR3 signaling activates caspase-8 through the FADD pathway in epithelial cell lines and fibroblasts (11), which can induce apoptosis (30). Although IL-1 secretion is classically associated with caspase-1 activation, some studies show that caspase-8 can process IL-1 downstream of select receptors (31, 32). No reports to our knowledge have examined caspase-8–mediated IL-1 regulation downstream of DR3 or NOD2. Both DR3 and NOD2 stimulation activated caspase-8 in MDM at 15 min (Fig. 5C). In contrast to NOD2, DR3 stimulation did not activate caspase-1 (Fig. 5C). As such, following effective caspase-8 and caspase-1 knockdown (Fig. S9C), caspase-8, but not caspase-1, was required for DR3-induced early IL-1 secretion (Fig. 5D) and long-term (24 h) cytokine secretion (Fig. 5F). In contrast, both caspases were required for NOD2-induced early IL-1 secretion (Fig. 5E) and long-term cytokine secretion (Fig. 5G). Cell survival was intact (Fig. S9D). Both caspases were also required for NOD2-induced TACE activity and early soluble TNFSF15 (Fig. S9E). Caspase-3 or caspase-4 silencing (Fig. S9F) did not alter DR3- or NOD2-mediated early secreted IL-1 (Fig. 5 D and E). Moreover, autocrine IL-1 and TNFSF15 cooperated with each other, as IL-1R stimulation led to early soluble TNFSF15 (Fig. S9G), which contributed to IL-1–induced signaling (Fig. S9H) and long-term cytokine secretion (Fig. S9I). Taken together, TNFSF15 and IL-1 secretion, regulated posttranslationally through TACE and caspases, respectively, cooperate and are required for the amplification of PRR-induced cytokines.

We next examined whether TNFSF15:DR3 interactions activate caspase-8 to induce IL-1 in human intestinal myeloid cells. We observed low, but significant, IL-1 secretion following DR3 stimulation (Fig. 5H). DR3-induced IL-1 was completely attenuated upon caspase-8, but not caspase-1, inhibition (Fig. 5H). S. typhimurium-induced IL-1 in intestinal myeloid cells amplifies IL-8 secretion (22). Consistently, caspase-8 inhibition attenuated not only IL-1, but also IL-8 secretion in TNFSF15-treated intestinal myeloid cells at 24 h (Fig. 5H). Furthermore, consistent with the poor response of intestinal macrophages to PRR ligands (19, 33), these cells did not secrete IL-1 or IL-8 following MDP treatment (Fig. 5H). Therefore, intestinal myeloid cells particularly depend on DR3 signaling for IL-1 secretion and therefore secretion of additional IL-1-dependent cytokines, thereby highlighting the importance of TNFSF15:DR3 interactions in these cells.

DR3-induced TRADD and FADD activate caspases in human epithelial cell lines and mouse fibroblasts (30, 34) (Fig. 5I). Moreover, dectin-1–initiated caspase-8 activation and IL-1 processing in human dendritic cells requires MALT1 (32). TRADD, FADD, and MALT1 were each required for DR3-induced early IL-1 secretion (Fig. S9J and Fig. 5J) and long-term secretion of additional cytokines (Fig. 5K). TNFSF15:DR3 signaling can proceed through another arm by using TRAF2 to activate NF-κB signaling in T cells (34) (Fig. 5I). Furthermore, signaling and outcomes via receptor-interacting protein (RIP) 1 and RIP3 downstream of TNF-superfamily receptors has been controversial and likely dependent on the stimulus and cell type (35, 36). Upon silencing TRAF2, RIP1, or RIP3 (Fig. S9J), we found that each of these molecules was dispensable for early DR3-initiated IL-1 secretion (Fig. 5J). However, each was required for long-term cytokine secretion (Fig. 5K), consistent with TRAF2 participating in a DR3-dependent NF-κB pathway (34); NF-κB mediates cytokine transcription. Therefore, TRADD/FADD/MALT1, but not TRAF2/RIP1/RIP3, signaling is required for TNFSF15:DR3-induced early IL-1; both pathways are ultimately required for DR3-induced cytokines through distinct mechanisms.

MDM from rs6478108 A Relative to GG Carriers Show Increased TNFSF15 Expression and NOD2- and DR3-Initiated Signaling.

We next questioned the mechanism through which rs6478108 regulates cytokine secretion. Rs6478108 is located in an intronic region of TNFSF15; we hypothesized that it might regulate TNFSF15 expression. Higher TNFSF15 expression in A risk carriers would increase amplification of PRR-induced cytokines. Consistent with Fig. 1 and Fig. S1, in untreated and MDP-treated MDM, TNFSF15 mRNA (Fig. 6A) and surface protein (Fig. 6B) expression was higher in rs6478108 A than in GG carriers. NOD2-induced soluble TNFSF15 was similarly increased in rs6478108 A compared with GG carrier MDM (Fig. 6C). Furthermore, ERK (Fig. 6D), p38, and NF-κB (Fig. S10A) activation was increased following MDP or TNFSF15 treatment, alone or in combination, of MDM from rs6478108 AA compared with GG carriers, consistent with increased TNFSF15 levels amplifying NOD2 and DR3 signaling. ERK is required for NOD2- and TNFSF15-induced cytokine induction (Fig. S5B) (22). Consistently, ERK inhibition reduced cytokine secretion in all three rs6478108 genotype carriers (Fig. S10B). TACE and TIMP3 expression were not affected by rs6478108 genotype (Fig. S10 C and D). In contrast, there was a small, but significant increase in early NOD2-induced TACE activity in rs6478108 AA carriers (Fig. S10E), consistent with their increased TNFSF15 expression, and TNFSF15 regulation of early TACE activation (Fig. S7H). Therefore, risk rs6478108 A carrier MDM show increased surface and soluble TNFSF15 and increased PRR- and DR3-induced signaling.

Fig. 6.

MDM from rs6478108 A risk carriers express increased TNFSF15 mRNA and protein and show increased NOD2- and TNFSF15-mediated ERK activation relative to GG carriers. (A–C) MDM from rs6478108 AA, GA, or GG carriers (n = 10, n = 10 and n = 10, respectively) were left untreated or treated with 100 μg/mL MDP for 4 h (A) or 24 h (B and C). Summarized data are represented as TNFSF15 mRNA expression (expressed as the change in CT values normalized to GAPDH and represented as a linear scale) (A), TNFSF15 protein expression (B), representative flow cytometry (B, Left), and summary of TNFSF15 surface protein (B, Right), or TNFSF15 in supernatants (C). (D) rs6478108 AA or GG carriers (n = 8 and n = 9, respectively) were treated for 15 min with 1 μg/mL MDP or 10 ng/mL TNFSF15, alone or in combination. (D, Left) Representative flow cytometry of phospho-ERK with MFI values. (D, Right) Summarized data for fold phospho-ERK induction normalized to untreated cells + SEM. Tx, treatment, T15, TNFSF15. Isotype controls, gray shading; untreated cells, dotted line; treated cells, solid line. *P < 0.05; **P < 0.01; ***P < 0.001; †, P < 1 × 10⁻⁴; ††, P < 1 × 10⁻⁵.

Discussion

Proper cytokine balance upon microbial exposure is essential for immune homeostasis; dysregulation is associated with immune-mediated diseases (1). Through independent knockdown and antibody blockade approaches, we demonstrate that TNFSF15:DR3 interactions dramatically enhance cytokine secretion induced by PRR ligands, mycobacterial components, and live bacteria. DR3 stimulation with TNFSF15, rather than vice versa, mediates these effects. TACE cleaves transmembrane TNFSF15 to soluble TNFSF15, and soluble TNFSF15 is sufficient to rescue decreased cytokine induction in MDM upon TACE inhibition. TNFSF15:DR3-mediated cytokine amplification requires TRADD/FADD/MALT-1/caspase-8–dependent, but caspase-1–independent, autocrine IL-1. Importantly, TNFSF15:DR3 signaling amplifies bacterial-induced cytokines in human intestinal myeloid cells ex vivo, and caspase-8 is required for DR3-induced IL-1 in these cells. Human myeloid-derived cells carrying the IBD- and leprosy-associated TNFSF15 region rs6478108 A allele express increased TNFSF15 and demonstrate increased PRR-induced signaling and cytokine secretion. Although rs6478108 GG carrier MDM shows only partial reduction of PRR-induced cytokine secretion because of a relative decrease in TNFSF15 expression compared with A risk carriers, complete TNFSF15 expression knockdown dramatically decreases cytokine secretion even at high PRR ligand doses. Therefore, we establish DR3 expression in human MDM, define a previously unidentified role for TNFSF15:DR3 in amplifying PRR-induced cytokines in MDM, elucidate TNFSF15:DR3 signaling mechanisms, and determine that disease-risk rs6478108 A carriers demonstrate a gain of function with increased PRR-initiated signaling and cytokines (Fig. S10F).

TNFSF15:DR3 signaling exacerbates inflammatory diseases in mice, including experimental colitis (10) and arthritis (37). TNFSF15 is elevated in intestinal tissues and serum of IBD patients (38, 39). That the TNFSF15 region rs6478108 A risk allele is a gain of function is consistent with these human and mouse findings. The rs6478108 A allele is the major allele, such that the majority of individuals are disease-risk carriers. Targeting of TNF family members has been a mainstay in immune-mediated diseases, and modulating TNFSF15:DR3 interactions is being considered for select inflammatory diseases. TNFSF15 contributions to intestinal inflammation have been attributed to TNFSF15:DR3 T-cell interactions, which increase T-cell activation and cytokine secretion. Our study highlights an additional, previously unknown, mechanism through which TNFSF15 can mediate intestinal inflammation, namely by enhancing PRR-mediated cytokine secretion in myeloid cells, and further highlights the potential for this pathway in therapeutic interventions.

Materials and Methods

Patient Recruitment and Genotyping.

Informed consent was obtained per protocol approved by the institutional review board at Yale University. Participants had no personal or family history of autoimmune/inflammatory disease, including psoriasis, systemic lupus erythematosus, rheumatoid arthritis, multiple sclerosis, type I diabetes mellitus, Crohn disease, and ulcerative colitis, or a history of HIV. Given the limitation in peripheral cell numbers and the range of innate responses we sought to examine, two separate cohorts of 100 and 98 European ancestry individuals were recruited for NOD2/TLR2 dose–response studies in MDM and MDDC, and NOD2/TLR synergy studies in MDDC, respectively. We genotyped polymorphisms by TaqMan SNP genotyping (Applied Biosystems) or Sequenom platform (Sequenom).

Primary Myeloid Cell Culture.

Monocytes were purified from human peripheral blood mononuclear cells by positive CD14 selection (Miltenyi Biotec), tested for purity, and cultured with M-CSF (10 ng/mL) (R&D Systems), or IL-4 (40 ng/mL) and GM-CSF (40 ng/mL) (R&D Systems) for 7 d for MDM and MDDC differentiation, respectively. Myeloid cells (CD11c purity >75%) were isolated as in ref. 19 from colonic resection specimens from uninvolved intestine in four non-IBD patients undergoing surgery for diverticular disease or colon cancer.

Myeloid Cell Stimulation.

Myeloid-derived cells were treated with MDP (Bachem), Pam₃Cys-Ser-(Lys)4 (EMD Millipore), lipid A (Peptides International), flagellin, CL097, CpG, poly I:C (Invivogen), DR3 (R&D Systems), TNFSF15 (Peprotech), or S. typhimurium at multiplicity of infection (MOI) 10:1. For antibody and inhibitor treatments, cells were incubated with anti-DR3 (eBioscience) antibody, IL-1Ra (IL-1R antagonist) (Genscript), TAPI-1, Ac-YVAD-Cho (YVAD), or caspase 8 inhibitor II (EMD Millipore) 1h before treatment. Supernatants were assayed for TNF-α, IL-8, IL-10 (BD Biosciences), IL-12, IL-23, IL-1β (eBioscience), or TNFSF15 (Peprotech) by ELISA.

Transfection of Small Interfering RNAs and Plasmids.

One hundred nanomolar scrambled or ON-TARGETplus SMARTpool small interfering RNA (siRNA) against TNFSF15, DR3, caspase-1, caspase-8, caspase-3, caspase-4, TRADD, FADD, TRAF2, RIP1, RIP3, or MALT1 siRNA (Dharmacon) were transfected into MDM by Amaxa nucleofector (Amaxa).

Protein Expression Analysis.

Proteins were detected by flow cytometry with phycoerythrin-labeled anti–phospho-ERK (Cell Signaling) or phycoerythrin-labeled anti-DR3 (eBioscience). Western blot (19) used anti-caspase-1, anti-caspase-8, (Cell Signaling), or anti-GAPDH antibodies (EMD Millipore).

mRNA Expression Analysis.

RNA was isolated, reverse transcribed, and quantitative PCR performed as in ref. 17. Each sample was run in duplicate and normalized to GAPDH. Primers sequences are available upon request.

Statistical Analysis.

Significance was assessed by using two-tailed t test. P < 0.05 was considered significant. The effects of polymorphisms on NOD2-induced cytokine secretion were analyzed by Mann–Whitney u test.