The TNF-family cytokine TL1A: from lymphocyte costimulator to disease co-conspirator

Review on the TNF-family cytokine TL1A in physiological and pathological immune responses.

Abstract

Originally described in 2002 as a T cell-costimulatory cytokine, the tumor necrosis factor family member TNF-like factor 1A (TL1A), encoded by the TNFSF15 gene, has since been found to affect multiple cell lineages through its receptor, death receptor 3 (DR3, encoded by TNFRSF25) with distinct cell-type effects. Genetic deficiency or blockade of TL1A-DR3 has defined a number of disease states that depend on this cytokine-receptor pair, whereas excess TL1A leads to allergic gastrointestinal inflammation through stimulation of group 2 innate lymphoid cells. Noncoding variants in the TL1A locus are associated with susceptibility to inflammatory bowel disease and leprosy, predicting that the level of TL1A expression may influence host defense and the development of autoimmune and inflammatory diseases.

Introduction

The TNF family of cytokines constitutes a large group of structurally and functionally related ligands and receptors that exert powerful effects on both innate and adaptive immune responses. TNF cytokines can costimulate T cells, B cells, ILCs, and osteoclasts, as well as induce programmed cell death. These functions are not exclusive, and one cytokine can have multiple effects, either signaling through the same receptor or through multiple receptors [1]. TL1A and its receptor DR3 fall into the subgroup of TNF-family cytokines that costimulate T cells, a property defined by the ability of these cytokines to enhance proliferation, survival, and cytokine production by T cells, independent of the TCR or the CD28 family of costimulatory receptors. TL1A and DR3 was one of the last TNF-family cytokine ligand-receptor pairs to be identified. DR3 was identified in the mid 1990s by a number of groups as a TNF-family receptor with high homology to TNFR1 but primarily expressed by lymphocytes and was given a number of names in addition to DR3 (TR3, DDR3, LARD, APO-3, TRAMP, WSL-1, WSLLR) [2–6]. Early studies identified the TNF-family cytokine TWEAK (TNFSF12) as a ligand for DR3, but this has not been confirmed, and the authentic receptor for TWEAK, Fn14 (TNFRSF12A) was subsequently identified.

The gene encoding the ligand for DR3 was originally described as encoding the transcript TL1, also named vascular-endothelial growth inhibitor because of its ability to induce endothelial cell apoptosis and inhibit angiogenesis [7, 8]. In 2002, an alternative, longer mRNA transcript originating from the same gene was identified by a group at Human Genome Sciences (Rockville, MD, USA). This transcript was named TL1A and was identified as the ligand for DR3 [9, 10]. TL1A encodes an N-terminal transmembrane domain, a metalloprotease cleavage site, and a C-terminal TNF homology domain more typical of other TNF superfamily members, which are synthesized as type II transmembrane proteins. TL1A transcripts are more abundant than TL1 transcripts and are likely the predominant form of the cytokine.

In addition to DR3, the original description of TL1A also identified DcR3, also called TR6 and TNFSF6B, as a receptor for this new ligand [9]. Along with TL1A, DcR3 also binds the TNF superfamily members LIGHT (TNFSF14) [11] and Fas ligand (TNFSF6) [12]. Although DcR3 is expressed across a range of human tissues and is overexpressed in many tumors [13], no homolog of this receptor has been identified in the mouse [14]. DcR3 is composed of 4 N-terminal, cysteine-rich domains and a C-terminal, heparin-binding domain. This receptor binds its multiple ligands via their conserved backbone, and the heparin binding domain enables interaction with multiple effector cell types [15]. Although the absence of DcR3 in the mouse genome has hampered understanding of its physiologic role, human DcR3 binds to mouse TL1A, LIGHT, and Fas ligand, and expression of DcR3 in transgenic mice blocks TL1A, LIGHT, and Fas ligand, suggesting that it functions as a negative regulator of TL1A-DR3 signaling in humans [16, 17].

Despite its moniker as “death receptor 3”, DR3 signaling rarely directly triggers cell death. This apparent paradox is rooted in the evolving understanding of TNF-receptor signaling. TNF-family receptors can be divided into those with and those without an intracellular death domain, which allows recruitment of adapter proteins with a death domain through homotypic interactions. Upon ligation by their cognate ligands, TNF-family receptors with a death domain, such as Fas and TRAIL receptors, recruit the adaptor Fas-associated death domain (FADD) protein, which, in turn, binds to caspase-8 (and caspase-10 in humans) through a structurally related death-effector domain. Upon recruitment to the signaling complex, caspase-8 can autocatalytically activate its enzymatic protease function and initiate a chain reaction of proteolysis, which, in turn, activates the cell death program through mitochondrial permeabilization and activation of the effector caspases-3 and -7. TNF receptors without a death-domain signal through recruitment of TRAF proteins via short peptide sequences in their cytoplasmic tails. TRAF proteins can act as E3 ubiquitin-ligases, ubiquitinating key substrates, such as RIP1, TAB2, and inhibitor of apoptosis proteins. Ubiquitination of these proteins allows recruitment and activation of kinase complexes, such as the IKK complex, which activates NF-κB transcription factors and MAPK, which activate the transcription factor AP-1 [18, 19].

Like its closest homolog TNFR1, DR3 primarily signals through TNFR-associated death domain (TRADD) protein. This adapter protein contains a death domain and a TRAF-binding domain, thus enabling TNFR1 and DR3 signaling to primarily activate NF-κB and MAPK [20]. Upon blockade of NF-κB activation, antiapoptotic NF-κB target genes are not made, and TNFR1 can convert to an apoptosis-inducing receptor through delayed secondary recruitment of Fas-associated death domain protein and caspase-8 [21]. Further blockade of caspase activation results in diversion of TNF signaling to necrotic cell death, which may enhance inflammation through release of mitochondrial DNA and other intracellular proinflammatory mediators [22]. DR3 signaling can also be diverted to cell death through inhibition of NF-κB in cell lines [2], but most of its effects in immune cells described thus far are thought to result from nonapoptotic pathways.

DR3 AND TL1A EXPRESSION AND FUNCTION IN DIVERSE CELL TYPES

Signaling through DR3 is controlled primarily by the availability of TL1A because TL1A expression is very low at baseline in most cell types and is highly inducible by proinflammatory stimuli (Table 1). However, expression of DR3 is also limited to certain tissues and can be induced to some degree (Fig. 1 and Table 2). Alternative splicing of DR3 mRNA adds another layer of regulation to DR3 signaling.

TABLE 1.

Major cell types expressing TL1A

Cell type or organ	Comments	References
T cells	Up-regulated on activation via the TCR on both CD4+ and CD8+ T cells in vitro	[23, 24]
F4/80+ macrophages	In response to chronic infection with Salmonella enterica serovar Typhimurium	[25]
Monocytes	Up-regulation upon stimulation with immune complex and TLR ligands	[26, 27]
Dendritic cells	Up-regulation upon stimulation with immune complex and TLR ligands	[23, 26, 27]
CX3CR1+ mononuclear phagocytes	High mRNA expression in cells isolated from murine colon	[28]
HUVEC cells	Highly expressed and inducible in response to IL-1α and PMA stimulation	[9]
Kidney vascular endothelial cells	mRNA and protein detected	[29]
Kidney tubular epithelial cells	Protein but no mRNA in allograft rejection	[29]
Murine brain	mRNA detected	[30]

Immune and nonimmune cells that express the ligand TL1A with their expression patterns and the conditions under which expression has been observed. HUVEC, human umbilical vein endothelial cells.

Figure 1. Effects of TL1A across leukocyte subsets.

Effects of TL1A-DR3 stimulation on various immune cell types are detailed in the text. DR3 ligation in T cells promotes proliferation and IL-2 production; enhances cytokine production, often in synergy with other stimuli; promotes Th9 differentiation while inhibiting iT_reg differentiation; and may affect Th17 differentiation, although results are study-dependent. By contrast, TL1A has been reported to inhibit B cell proliferation. In ILCs, TL1A enhances signature cytokine production in synergy with canonical ILC subset stimuli. Effects of TL1A-DR3 signaling on myeloid lineage cells are less well understood, but there is evidence that TL1A synergistically enhances macrophage pattern recognition receptor (PRR) signaling, as well as foam cell and osteoclast differentiation.

TABLE 2.

Major cell types expressing DR3

Cell type or organ	Comments	References
IgM+ B cells	Increased upon activation with anti-IgM	[31]
B220− CD138− IgM- plasma cells	15 d after BTIIC/CFA immunization	[32]
CD4+ T cells	Increases after activation; Th17 and Th9 show greater expression than Th0/1/2	[33–35]
CD8+ T cells	Transient increase after activation	[33, 36]
CD161+ CD4/8+ T cells	Subpopulation of DR3+, which increases with low IL-12/IL-18 stimulation	[37]
FoxP3+ CD4+ T cells	Ex vivo Treg, Foxp3-GFP reporter mouse: CD4+, GFP+, and GFP− express similar levels ex vivo	[38, 39]
	iTreg: comparable to Th17	[34, 35]
NK cells/ILC1	Ex vivo, lower than CD4+ T cells from mesenteric lymph nodes	[24, 40]
	Subpopulation of DR3+ among NK/ILC1 cells	[41]
	NK1.1+ CD3−
	Lin− NK1.1+
NKT cells	Ex vivo
	NK1.1+ TCRβ+	[33]
	NK1.1+ CD3+	[24]
ILC2	Ex vivo, comparable to CD4+ T cells from mesenteric lymph nodes
	Lin− NK1.1− KLRG1+	[41]
	Mouse: Lin− Sca1+ CD127+ ST2+	[42]
	Human: Lin− CRTH2+ CD161+ CD127+
ILC3	Ex vivo, comparable to CD4+ T cells from mesenteric lymph nodes	[41]
	Lin− NK1.1− KLRG1− RORϒt+ CD127+
Monocyte-derived macrophages	M-CSF cultured	[43]
CD11c+ cells	Subpopulation of DR3+ cells directly ex vivo	[24]
Kidney tubular epithelial cells	Normal human kidney and increased in ischemic allographs	[44]
Glomerular endothelial cells	Ischemic allographs; not in healthy tissue.	[44]
Brain neuronal linage cells	Neurons within hippocampal and cortical regions	[30]

Immune and nonimmune cells that express the receptor DR3 with their expression patterns and the conditions under which they have been observed. Where expression has been demonstrated on multiple cell types within a publication, approximate comparisons between relative expression levels have been given. Where different groups have used alternative methods to identify the same population, the markers used have been noted. BTIIC, bovine type II collagen; CRTH2, chemoattractant receptor homologous molecule expressed on T helper Type 2 cells; GFP, green fluorescent protein; KLRG1, killer cell lectin-like receptor subfamily G, member 1; Lin, lineage markers; M-CSF, macrophage colony-stimulating factor; RORϒt, RAR-related orphan receptor gamma t; Sca1, stem cell antigen 1.

Originally, 11 splice variants were described in human DR3; however, some of these contain intronic sequences and may be cloning artifacts [2]. Of the splice variants known to be expressed, variant 1 is the full-length isoform of the protein containing 4 extracellular cysteine-rich repeats, a transmembrane domain, and an intracellular death domain. Variant 3 lacks exons 5 and 6, which lead to an absence of cysteine-rich repeat 4, and variant 2 lacks exon 6, which introduces a frame-shift mutation resulting in a truncation that deletes the transmembrane and death domains, encoding a putative, soluble form of the receptor [33, 45]. T cell activation tends to favor the full-length variant [33, 34], which may serve to enhance DR3 signaling by activated T cells. However, Foxp3⁺ T_regs express a greater proportion of variant 3 [34].

Lymphocytes

There is considerable variation in DR3 expression within the T cell compartment, but it is detectable on almost all subsets (Table 2). Within mouse T cells, memory (CD44^hi) cells have higher DR3 expression compared to naïve cells. DR3 is up-regulated in response to stimulation via the TCR, with expression peaking at 3 d in CD4⁺ T cells. In CD8⁺ T cells, DR3 expression peaks much earlier, at day 1, and, by day 4, declines to unstimulated levels of expression [33]. DR3 expression further differs between functional CD4⁺ T cell subsets, with Th9, Th17, and T_reg expressing higher surface DR3 compared with Th0, Th1, and Th2 subsets [34, 35]. However, DR3 expression levels are equivalent on CD4⁺ Foxp3⁻ conventional T cells and CD4⁺ Foxp3⁺ regulatory T cells taken from a Foxp3 reporter mouse [38].

Although TL1A was originally characterized in endothelial cells, its expression by a variety of leukocyte subsets was quickly identified [46]. Similar to DR3, TL1A is induced on T cells following activation, with minimal expression in naïve T cells (Table 1). Autocrine feedback via DR3 may help control TL1A levels because DR3-deficient T cells have reduced induction of TL1A in response to TCR signaling [23, 24].

TL1A and DR3 have recently been found expressed by other lymphocyte subsets (Fig. 1 and Tables 1 and 2). DR3 is up-regulated as a result of IgM stimulation of human B cells, with greatest expression before class switching, whereas TL1A expression is not detected [31]. Following immunization of mice with antigen, DR3 expression is observed on class-switched B220⁻ CD138⁺ plasma cells but not other class-switched B cells or plasmablasts [32]. In vitro, the addition of TL1A can inhibit proliferation of human B cells [31], but the physiologic function of DR3 on B cells is not yet clear. A subpopulation of resting NK cells has also been shown to express DR3 [24]. After incubation with IL-12 and IL-18 for 48 h, most cells express DR3, and the addition of recombinant TL1A increases IFN-γ production and cytotoxicity [40, 47]. NKT cells exhibit moderate constitutive expression of DR3 that is greater than that observed on ex vivo CD4⁺ T cells taken from a murine lymph node [24].

In recent years, TL1A has been shown to have an effect on innate lymphoid cells, a recently discovered population that develops from common lymphoid progenitors but lacks T or B cell or other specific lineage markers [48]. ILCs are generally tissue-resident cells that can secrete large amounts of cytokines in response to stimuli. To date, these stimuli have been mainly identified as other inducer cytokines. ILCs are considered the innate counterpart of Th1, Th2, or Th17 T cell subsets and are grouped accordingly as ILC1, ILC2, and ILC3, depending on the cytokines they produce [48, 49]. ILCs isolated from the mesenteric lymph nodes or the intestinal lamina propria express similar levels of DR3 as CD4⁺ T cells isolated from the same tissues (Table 2). Upon close examination, ILC2 (Lin⁻/NK1.1⁻/KLRG1⁺) and ILC3 (Lin⁻/NK1.1⁻/KLRG1⁻/RORϒt⁺/CD127⁺) express higher levels of DR3 than NK cells/ILC1 (Lin⁻/NK1.1⁺) [41, 42]. In addition, DR3 expression can be detected on human ILC2 isolated from peripheral blood [42].

Myeloid cells

Myeloid cells can be induced to express large amounts of TL1A (Table 1). Within a few hours of stimulation, monocytes, macrophages, and dendritic cells up-regulate TL1A gene expression and produce large quantities of protein. TL1A-inducing stimuli include TLR ligands, with the exception of single-stranded RNA ligands for TLR7 and 8, and immune complexes, which trigger signaling through FcRs [23, 26, 27, 50]. In vivo, infection with Salmonella typhimurium induces splenic F4/80⁺ macrophages to express TL1A in situ, which may be important for host defense against Salmonella as DR3-deficient mice have reduced clearance of the bacteria [25, 27]. Similarly, TL1A mRNA is highly expressed in CX3CR1⁺ mononuclear phagocytes, a population that has a sentinel role in the intestinal lamina propria responding to microbial products [28]. Although the role of DR3 on myeloid cells has been examined considerably less (Table 2), a recent study demonstrated that human monocyte-derived macrophages express DR3 and that DR3 signaling synergistically enhances NOD2 and other pattern recognition receptor signaling in these cells via autocrine TL1A and IL-1 [43] (Fig. 1). In macrophages, TL1A can also promote uptake of oxidized LDL, metalloproteinase expression, and foam cell differentiation [51, 52], implicating TL1A-DR3 interactions in the pathogenesis of atherosclerosis. TL1A has been shown to promote osteoclast differentiation in vitro [53], which may have a role in the effects of TL1A on joint damage in arthritis.

Expression of DR3 and TL1A outside of the immune system

TL1A can also be expressed outside the immune system (Table 1). Endothelial cells can produce TL1A, with high levels of expression in human umbilical vein endothelial cells inducible with PMA or IL-1α [9]. The relative contributions of endothelial and myeloid cells to elevated serum levels of TL1A seen in inflammatory states, such as RA, are not known. Both DR3 and TL1A have been shown to be expressed in the kidney (Tables 1 and 2), particularly in the setting of tubular injury or graft rejection, suggesting that TL1A-DR3 interactions may mediate renal pathology independent of the immune system. DR3 is expressed in scattered endothelial cells in the normal human kidney, but, in the setting of allograft rejection, DR3 expression is up-regulated in the glomeruli, endothelial cells, and interstitium, in addition to infiltrating immune cells [44]. In terms of the ligand, kidney vascular endothelial cells express TL1A mRNA and protein in both healthy and diseased tissue. However, TL1A protein, but not mRNA, has been observed in the tubular epithelial cells in cases of rejection-mediated damage, suggesting that this is a result of uptake of TL1A, which may then be able to interact with DR3 present at this site [29].

Interestingly, DR3 and, to a lesser extent, TL1A are expressed in neurons, particularly the cerebral cortex, hippocampus, and dentate gyrus [30] (Tables 1 and 2). DR3-deficient mice develop an age-dependent loss of motor control manifested by gait and behavioral disturbances, which are associated with loss of cortical innervation of the striatum [30]. This suggests that tonic TL1A-DR3 interactions may be necessary to sustain survival of these motor neurons. Although TL1A is not the only cytokine in the TNF family to be implicated in neuronal function, it is unique in that such a striking neurologic phenotype has been found in mice lacking its receptor. Whether TL1A has a role in nervous system repair after immune injury, as has been shown for TNF [54], is not known.

T CELL COSTIMULATION AND DIFFERENTIATION INDUCED BY TL1A

The effects of TL1A on T cells differ between functional subsets (Fig. 1), but the most pervasive outcome of TL1A costimulation is promotion of IL-2 signaling. During T cell activation, the addition of TL1A increases IL-2 production and induces IL-2RA (CD25) and IL-2RB expression [9, 23, 34, 36, 55–58]. Functionally, TL1A increases T cell proliferation, particularly under suboptimal conditions in a manner partly dependent on IL-2 [9, 23, 36]. Defects in the proliferative responses of DR3-deficient T cells are more subtle and depend on suboptimal stimulation in the presence of antigen-presenting cells, suggesting that TL1A production by T cells is not sufficient to costimulate T cell activation [23].

Early studies of TL1A highlighted its ability to promote Th1-associated cytokines. Addition of TL1A to TCR-stimulated T cells enhances IFN-γ production particularly under conditions of suboptimal TCR activation [9, 59, 60]. TL1A also increases IFN-γ, in addition to a number of other proinflammatory cytokines, in synergy with IL-12 and IL-18 stimulation [40, 55, 61]. However, blockade of IL-12 and IL-18 does not prevent TL1A-induced enhancement of IFN-γ [59], indicating that TL1A does not simply up-regulate these Th1-promoting cytokines. A number of reports have examined the specific CD4⁺ T cells that respond to TL1A in synergy with IL-18 and IL-12, identifying CCR9 [60], CD161 [37, 55], and IL-18Rα [61] as markers of the T cell subsets that up-regulate DR3 upon IL-12 and IL-18 stimulation and produce IFN-γ in response to TL1A. Intriguingly, these T cell subsets are often found in the intestinal mucosa, suggesting a mechanism for enhancement of TL1A costimulation in the gut. In accordance with in vitro data, reduced levels of IFN-γ are seen in mouse models of Th1-dependent immune responses, such as Salmonella infection [25], and in EAE, which depends on both Th1 and Th17 cells [23]. Despite these demonstrations of TL1A-driven IFN-γ production, DR3-deficient T cells show no defect in Th1 differentiation [23], indicating that although TL1A may enhance cytokine production, it is not required for Th1 polarization.

The role of TL1A in enhancing immune responses is not, however, confined to IFN-γ. Although human cells reportedly do not produce Th2 cytokines in response to TL1A costimulation [9, 40], the addition of TL1A or agonistic anti-DR3 antibodies during murine CD4⁺ T cell activation increases the secretion of IL-4 and other Th2 cytokines [23, 24]. In a mouse model of lung hypersensitivity induced by the transfer of ovalbumin-specific Th2 cells and airway ovalbumin challenge, DR3-deficient T cells exhibit reduced lung expansion and induce milder disease [23], demonstrating that endogenous TL1A-DR3 can promote Th2-mediated immunopathology. However, as with Th1, TL1A-DR3 signaling is not necessary for Th2 differentiation of naïve murine CD4⁺ T cells [23].

The effects of TL1A on Th17 cells are complex, likely because of the competing effects of TL1A and TL1A-induced IL-2 on Th17 differentiation. DR3 expression is highly up-regulated on Th17 cells [34, 58], although Th17 differentiation is not impaired in DR3-deficient T cells [23]. TL1A promotes Th17 proliferation [34, 58], implying a positive role for this cytokine in Th17 biology. Pappu et al. [34] demonstrated that TL1A promotes in vitro Th17 differentiation, but only in the presence of IL-2 inhibition, as might be expected, since IL-2-STAT5 signaling inhibits Th17 polarization [62]. By contrast, Jones et al. [58] reported that TL1A inhibits human and murine Th17 differentiation, independent of IL-2 signaling status. Subtle differences in TGF-β concentration or plate-bound vs. soluble anti-CD28 may be responsible for this discrepancy. TL1A and IL-23 can also enhance IL-17 production in CD4⁺ T cells extracted from both healthy and inflamed lamina propria [63, 64], indicating a second role for TL1A in enhancing IL-17 production in T cells already differentiated into IL-17 producers. In a T cell transfer model of EAE, both DR3-deficient T cells and T cells transferred into TL1A-deficient hosts expand less in the CNS and cause less-severe disease [23, 34], suggesting that endogenous TL1A-DR3 interactions are important in the pathogenesis of this IL-17-dependent model.

TL1A has also been reported to influence the generation and function of T_regs in a number of ways. Stimulation by soluble TL1A or anti-DR3 agonistic antibody increases T_reg proliferation both in vitro [65] and in vivo [39]. As with Th17, some of these effects may be mediated by IL-2. Despite this expansion of T_reg numbers, TL1A dampens the suppressive capacity of T_regs in in vitro assays, largely because of the ability of TL1A to costimulate the responder T cells [65, 66] but also potentially because of a direct effect by TL1A on the T_reg cells [65]. In contrast to effects on proliferation, during iT_reg differentiation from naïve T cell precursors in the presence of TGF-β and IL-2, TL1A actually inhibits the production of Foxp3⁺ cells [35, 66]. Alterations in the numbers and functions of T_regs in DR3-deficient mice have not, to our knowledge, been reported, and the role of endogenous TL1A-DR3 interactions on T_regs remains to be defined.

The negative effects of TL1A on iT_reg generation raise the question of whether TL1A might divert T_reg cells toward another fate. Indeed, Richard et al. [35] recently found that the addition of TL1A to iT_reg cultures promoted the generation of T cells that secrete IL-9 and lack Foxp3 expression. Although IL-9 production by T cells has been known for many years [67, 68], the status of IL-9-producing T cells as a separate lineage, Th9, which promotes allergic immune responses, has only recently become evident [69, 70]. Evidence that human T cells can also differentiate into predominantly IL-9 producers and that tissue from patients with allergic disease exhibits increased Th9 has prompted interest in the development and function of this subset [71–74]. Under conditions previously reported to optimize Th9 differentiation from naïve cells (TGF-β and IL-4), the addition of TL1A greatly enhances the generation of IL-9-producing cells. Surprisingly, IL-2 signaling through STAT5, rather than IL-4 through STAT6, is required for this effect [35]. Another TNF family cytokine, OX40L can also promote Th9 generation, but, in contrast to TL1A, the effect of OX40L is mediated by IL-4 and STAT6 [75], leaving TL1A unique in its ability to bypass the STAT6 axis for Th9 differentiation. Th9 pathogenicity during allergic inflammation is also enhanced by TL1A-DR3 signaling. In adoptive transfer models of both lung hypersensitivity and ocular inflammation, DR3-deficient, antigen-specific transgenic Th9 cells were less pathogenic than their wild-type counterparts [35]. Thus, TL1A-DR3 interactions constitute a novel pathway that promotes Th9 differentiation and pathogenicity, suggesting that preventing TL1A expression or signaling in the setting of allergic disease may be beneficial.

Although not as well studied as in CD4⁺ T cells, TL1A has been found to costimulate CD8⁺ T cells, both in primary [33, 36] and secondary responses [36]. However, in murine cytomegalovirus infection, DR3 signaling by CD8⁺ T cells is necessary only for mounting an appropriate primary response because DR3-deficient mice have a defect in initial CD8⁺ T cell proliferation and increased viral titers but, later, produce a memory response comparable to that of wild-type mice [33]. Tumors overexpressing TL1A are strongly rejected in a CD8-dependent manner [36], suggesting that TL1A costimulation is relevant to tumor immunity.

Effects of TL1A on NKT cells appear to be context-dependent. Papadakis et al. [60] demonstrated that TL1A stimulation of NKTs treated with IL-12 and IL-18 enhances IFN-γ expression. In a contrasting system, Fang et al. [24] showed that adding agonistic anti-DR3 antibody to TCR stimulation of NKT cells increases IL-13 expression and that, in a model of allergic lung disease, mice with NKT cells deficient in TL1A-DR3 signaling have dramatically lower type 2 cytokine expression in the lung. To what extent the NKT cell populations from these studies overlap with the TL1A-responsive CD4⁺ CD161⁺ T cell population [37, 55] remains to be determined.

EFFECTS OF TL1A ON INNATE LYMPHOID CELLS

Similar to T cells, TL1A has been reported to have a costimulatory effect on innate lymphoid cells, especially group 2 and 3 ILCs (Fig. 1). TL1A promotes production of the signature type 2 cytokines IL-5 and IL-13 in ILC2 [41, 42], and TL1A can synergize with IL-25 and IL-33 to promote human ILC2 cytokine expression and function [42]. Similarly, TL1A synergizes with IL-1β and IL-23 to induce IL-22 production by both human and mouse intestinal ILC3 [28]. Evidence that TL1A can costimulate ILCs, combined with the importance of ILC2-derived IL-13 in mounting an adaptive type 2 response [76], prompted 2 studies to examine the role of TL1A in costimulating ILC2 in allergic lung-disease models [41, 42]. Expansion of ILC2 is reduced in the inflamed lung of DR3-deficient mice during both ovalbumin-induced lung inflammation and protease allergen-driven lung pathology [41, 42], in addition to reduced numbers of Th2 and Th9 [23, 35]. This lack of expansion in the absence of DR3 appears to be an intrinsic ILC2 defect because Rag1-deficient mice that lack T and B cells exhibit similar results [42]. This highlights the concept that intact DR3 signaling pathways on both ILC2 and T cells contribute to pathology during lung inflammation.

Similar cooperativity between ILC2s and CD4⁺ T cells promotes expulsion of intestinal helminthic parasites [77]. Although chronic expression of TL1A has little effect on lung histology, it leads to spontaneous intestinal inflammation with pathology similar to that of mice infected with helminths [65, 66]. This intestinal pathology can be largely attributed to the substantial amount of IL-13 released by ILC2 in response to chronic stimulation by TL1A. Interestingly, TL1A transgenic mice have elevated levels of IL-25 and IL-33, epithelial cytokines triggered by parasitic infection, which can induce ILC2 to produce IL-5 and IL-13 [41]. TL1A can also directly stimulate ILC2 to produce IL-5 and IL-13 [42]. The similarity between the pathology developed by TL1A transgenic mice and that occurring in parasitic infection raised the question as to whether TL1A-DR3 signaling might have a role during helminth infection. Indeed, a recent study by Yu et al. [42] would suggest this because they found DR3-deficient mice to have impaired clearance of Nippostrongylus brasiliensis worms. However, Meylan et al. [41] performed similar experiments with a different line of DR3-deficient mice and observed normal kinetics of IL-13 production and parasite clearance during Nippostrongylus infection. One possible explanation for these contrasting results is that Yu et al. [42] treated mice with oral antibiotics for 5 d following parasite infection. Altered commensal flora in the setting of antibiotic treatment may have affected the immune response or growth of Nippostrongylus in the intestine. If so, this would imply that DR3 has a greater effect on intestinal parasite host defense in a more “sterile” environment. Whether TL1A is required for host defense against other intestinal parasites remains to be seen.

DR3-TL1A INTERACTIONS IN ANIMAL MODELS OF AUTOIMMUNE DISEASE

Despite the modest effects of DR3 deficiency on in vitro T cell activation and differentiation, a wide variety of T cell-dependent and T cell-independent animal models of autoimmune disease are dependent on TL1A and DR3, underscoring the importance of this cytokine-receptor pair in promoting autoimmune immunopathology. Rather than enhancing priming or expansion of antigen-specific pathogenic T cells, as do other members of the TNF superfamily, DR3 primarily promotes accumulation of pathogenic T cells in the target organ (Table 3).

TABLE 3.

Role of TL1A and DR3 in models of immune pathology vs. host defense

Pathology
Model	Intervention	Phenotype	References
EAE	DR3 KO	Reduced severity and cellular infiltration	[23]
	TL1A KO		[34]
Ovalbumin lung hypersensitivity	DR3 KO	Reduced severity and cellular infiltration	[35, 23]
	Antagonistic anti-TL1A (prophylactic and treatment)		[24]
Papain-induced allergic lung disease	DR3 KO	Reduced severity and cellular infiltration	[35, 41]
TNBS-induced colitis	Antagonistic anti-TL1A, DR3-FC, anti DR3 Fab (prophylactic)	Reduced pathology and mortality	[66]
DSS-induced colitis	Antagonistic anti-TL1A (prophylactic)	Reduced weight loss	[64]
	DR3 KO		[66]
Antigen induced arthritis	DR3 KO	Reduced joint pathology	[53, 78]
Collagen-induced arthritis	Antagonistic anti-TL1A (prophylactic)	Reduced pathology	[53]
	TL1A KO		[32]
Salmonella infection	DR3 KO	Reduced clearance of bacteria	[25]
MCMV and vaccinia virus infection	DR3 KO	Increased viral titers, increased morbidity and mortality	[33]

Findings and phenotypes from various murine models in which TL1A-DR3 signaling has been interrupted, either through genetic ablation of the receptor or ligand or via an antagonistic anti-TL1A monoclonal antibody treatment. Where more than 1 method of intervention is listed, the phenotype noted was observed in all methods. DSS, dextran sulfate sodium; KO, knockout; MCMV, murine cytomegalovirus; TNBS, 2,4,6-trinitrobenzenesulfonic acid.

The Samp1/YitFc and TNF^ΔARE models of IBD were 2 of the first spontaneous disease models in which TL1A and DR3 were found to be up-regulated [79]. DR3 deficiency or blocking TL1A with monoclonal antibodies also reduces disease in induced mouse models of IBD. DR3-deficient mice have a modest reduction in weight loss compared to control mice in the acute dextran sulfate sodium colitis model [66]. In chronic dextran sulfate sodium colitis, treatment with a blocking monoclonal antibody against TL1A effectively protects against the development of disease and attenuates preestablished disease by reducing activation of both Th1 and Th17 cells [64]. In the acute 2,4,6-trinitrobenzenesulfonic acid colitis model, which depends on T cells secreting IFN-γ, anti-TL1A antagonists significantly protect mice from weight loss, microscopic inflammation, and mortality [66].

TL1A-DR3 interactions also promote pathology in experimental models of arthritis. In antigen-induced arthritis, a local model of arthritis, DR3-deficient mice exhibit a reduction in joint pathology and notably fail to develop early cartilage destruction and subchondral bone erosion [53, 78]. Reduced clinical disease scores and joint pathology were observed in collagen-induced arthritis in mice treated with anti-TL1A antibodies [53] or mice genetically deficient in TL1A [32]. Conversely, injection of exogenous TL1A aggravates arthritis and elevates anticollagen antibody responses in this model [32, 80]. These data suggest that TL1A may be part of a feed-forward loop involving TNF and IL-17, which are pathogenic cytokines in human RA. TNF can induce monocytes and macrophage-like synovial cells to produce TL1A, which can, in turn, costimulate T cells to produce greater amounts of TNF and IL-17. Through such a loop, expression of these inflammatory cytokines may build up to pathogenic levels in a DR3-dependent manner [80, 81].

In models in which antigen-specific T cells can be tracked, an interesting dichotomy has emerged between systemic priming and expansion of pathogenic T cells and their accumulation in the target organ. DR3- and TL1A-deficient mice are relatively resistant to EAE and ovalbumin-induced airway inflammation, with reduced pathogenic cytokine production and accumulation of antigen-specific T cells in the target organ and draining lymph nodes. Despite these defects, priming of antigen-specific T cells in the spleen and lymph nodes is relatively normal [23, 24, 34]. Transfer experiments have demonstrated that some of this effect is intrinsic to T cells and may result from a lack of costimulation by TL1A during secondary stimulation of T cells in these disease models [23]. However in allergic lung disease, TL1A costimulation of ILC2 and NKT cells in the lungs is also critical for the generation of airway and parenchymal inflammation [24, 41].

TL1A IN HUMAN DISEASE

An essential question arising from studies that implicate TL1A in the pathogenesis of mouse models of autoimmunity is how relevant these findings are to human disease. Up-regulation of TL1A is characteristic of a variety of human inflammatory diseases (Table 4), particularly those sensitive to TNF inhibition, although whether it is reactive or pathogenic remains unknown. Increased TL1A expression was first observed in the inflammatory bowel diseases CD and UC [46]. TL1A is up-regulated in inflamed regions of the intestine in both CD and UC at the protein and mRNA levels [46, 55, 59, 82, 87], although mRNA expression in non-IBD colitis is also elevated [82]. Several studies have demonstrated elevated serum TL1A in patients with UC, both in those with active and those with quiescent diseases [87, 88] as well as in patients with active CD disease [82]. Interestingly, patients with CD and colonic involvement have elevated serum TL1A levels greater than those without such manifestations [82], suggesting a colonic bias for TL1A as a serum biomarker. Lamina propria macrophages, CD4⁺ T cells, and CD8⁺ T cells were found to express TL1A in patients with CD [46, 59, 63], whereas only lamina propria plasma cells stained positive for TL1A in patients with UC [46]. DR3 expression is also up-regulated on lamina propria T cells from inflamed intestinal tissue of individuals with CD or UC [46, 59]. Interestingly, intestinal TL1A mRNA expression is reduced in biopsy samples from patients with CD who are receiving anti-TNF or other immunomodulatory standard-of-care treatments compared with untreated patients [55].

TABLE 4.

TL1A associations with human autoreactive disease

Disease	Expression association	References	Genetic association	References
Crohn’s disease	Up-regulated mRNA and protein in inflamed intestine	[46, 55, 59, 82]	Japanese and European cohorts	[83–86]
	Up-regulated in serum in active disease
	Higher with colonic involvement.
	Lower mRNA in treated patients
Ulcerative colitis	Up-regulated mRNA and protein in inflamed intestine	[46, 55, 87, 88]	European cohorts	[86]
	Up-regulated in serum
Rheumatoid arthritis	Up-regulated in synovial fluid and serum.	[80, 89–92]
	Higher in severe/terminal disease.
	Serum TL1A decreases with anti-TNF treatment.
	Associated with RF and anti-CCP antibodies.
	Correlated with carotid plaque height progression.
Psoriasis	Up-regulated in serum.	[93, 94]
	Up-regulated protein and mRNA in skin lesions.
	Serum TL1A decreases with treatment.
Primary biliary cirrhosis	Up-regulated in serum in early and late stages.	[95]	Japanese cohort	[96]
	Higher in patients with cholangitis.
	Serum TL1A decreases with treatment in early stage.
Ankylosing spondylitis	Up-regulated in serum of patients not receiving anti-TNF treatment. Correlates with disease activity.	[97]

TL1A expression and genetic associations with human disease. Further details, particularly of genetic associations, are described in the text. References are given for each section.

In RA, increased soluble TL1A is found in both synovial fluid and serum [80, 89], particularly in patients with severe or terminal RA [90]. Elevated serum TL1A decreases after anti-TNF treatment, and levels in patients in remission appear no different from those in healthy controls [90], suggesting that TL1A is at least a marker of active inflammation in this disease. High levels of serum TL1A in RA are associated with increased rheumatoid factor and with anticyclic, citrullinated peptide antibodies [90, 91]. A recent report indicated that serum TL1A levels correlate with the rate of progression of carotid plaque height in a 3.5-y follow-up of patients with RA [92]. Although TL1A has been found on a variety of cell types that make up atherosclerotic plaques [98], it remains unclear whether TL1A has a direct role in their formation in patients with RA. Serum levels of TL1A in patients with RA strongly correlate with those in synovial fluid, although at lower concentrations [89]. These data demonstrate that serum TL1A is a good predictor of synovial fluid TL1A levels, suggesting that serum TL1A in RA may originate from the synovia rather than, for example, endothelial cells. Indeed, synovial macrophages from rheumatoid factor-positive RA patients express TL1A, potentially induced by immune complexes [81], and both synovial fibroblasts and chondrocytes are capable of producing TL1A upon stimulation by cytokines, including TNF [80].

Increased TL1A expression has recently been found in additional autoimmune and autoinflammatory diseases. Psoriasis patients exhibit elevated serum TL1A compared with atopic dermatitis patients or healthy controls, and this level decreases with treatment [94]. Intriguingly, TL1A is also elevated at the protein and mRNA level in skin lesions of patients with psoriasis, with immunohistochemical staining indicating TL1A production by a variety of immune (monocytes, neutrophils, lymphocytes) and nonimmune cells (keratinocytes, basal cells, endothelial cells) in patients but not in healthy controls [93]. Up-regulated serum TL1A was also recently demonstrated in a Japanese PBC cohort, although patients with chronic hepatitis C and autoimmune hepatitis show a similar increase [95]. Patients with PBC, in both early and late-stage disease, exhibit elevated TL1A expression, particularly patients with cholangitis, but serum TL1A levels were only found to decrease with treatment in the early stage group [95]. Staining of liver biopsies revealed constitutive TL1A expression in intrahepatic bile ducts, blood vessels, and Kupffer cells, with additional TL1A expression on infiltrating mononuclear cells from PBC and chronic hepatitis C patients [95]. Finally, patients with ankylosing spondylitis not treated with TNF-blocking therapies have increased serum TL1A compared with both healthy controls and patients receiving anti-TNF treatment (monoclonal antibody or recombinant TNFR-Fc fusion protein) [97]. Furthermore, serum TL1A in patients with ankylosing spondylitis was found to correlate with disease activity scores [97]. Taken together, these studies indicate that TL1A expression tends to correlate with disease activity in a variety of autoimmune and autoinflammatory settings, such that systemic levels are particularly increased in active disease and decrease with treatment. In these inflammatory contexts, TL1A is found on a variety of cell types, particularly at the site of lesions or immunological activity.

Expression of DcR3 is often up-regulated in the same contexts as TL1A, resulting in associations with RA [90–92], CD [82, 99], UC [88], psoriasis [93, 100], and PBC [95] incidence and/or activity. Interestingly, DcR3 expression is also elevated in autoimmune diseases in which no TL1A activity has been reported, including systemic lupus erythematosus [101, 102] and primary Sjögren’s syndrome [103]. Such associations may reflect the role of DcR3 as a decoy receptor for additional TNF family ligands or may indicate as yet undiscovered functions for TL1A in these contexts.

While TL1A expression appears to correlate with inflammatory activity, the question remains whether it is causal or a reaction to the disease state. One clue in the search for a causal role for TL1A in human disease comes from genetic association studies (Table 4). Genetic association of the TNFSF15 locus encoding TL1A with disease was first established in a case-control, genome-wide association study (GWAS) in 2005 in a Japanese cohort of patients with CD, with suggestions of a similar association in 2 UK cohorts [83]. TNFSF15 has remained the strongest non-HLA susceptibility locus for CD in the Japanese population in subsequent GWASs, although it has not been associated with UC in this population [84, 85]. Association of TNFSF15 with both CD and UC has been confirmed in numerous GWASs of populations of European descent, including a comprehensive meta-analysis [86]. GWASs of specific patient subsets, including patients with pediatric onset IBD vs. healthy controls [104] and patients with medically refractory UC vs. healthy controls [105], also report association at the TNFSF15 locus. Intriguingly, Picornell et al. [106] suggested that the CD risk and protective TNFSF15 haplotypes may be reversed in the Ashkenazi Jewish population, but this finding has not been replicated in subsequent studies [107, 108]. In a targeted analysis, the same IBD-risk variant was found associated with risk for irritable bowel syndrome, particularly irritable bowel syndrome with constipation [109]. In addition to TNFSF15, the genomic region near TNFRSF6B, which encodes DcR3, has also been associated with IBD [86], and a recent study demonstrated that patients with pediatric-onset IBD have a higher frequency of rare missense mutations in TNFRSF6B [110].

Genetic variation in the loci encoding TL1A and DR3 have been associated with other diseases, although less definitively than with IBD. A recent Japanese study associated the TNFSF15 locus with risk of another autoreactive disease, PBC [96], but unlike the IBD association, genome-wide association was not found in the most recent European PBC GWAS [111]. Additionally, larger cohorts will be necessary to confirm this difference, but should it hold, the population specificity suggests an interaction of TNFSF15 genetic elements with environmental factors or population-specific genetic variation. Despite the increased TL1A expression in RA, genetic variants near TNFSF15 have not been found to contribute to RA risk. The TNFRSF25 locus (encoding DR3) was associated with RA by early linkage studies [112, 113], and a targeted study in 2004 found TNFRSF25 copy-number variation, associating RA with increased copies [114], but subsequent GWASs, represented by references 115–117, have not identified genetic variants near TNFSF15 or TNFRSF25 associated with RA. Interestingly, the TNFSF15 locus was also found to be associated with leprosy in a GWAS of Chinese individuals [118, 119], adding to the list of IBD-associated loci that are shared with mycobacterial diseases. Along with evidence from mouse models of bacterial [25] and viral infection [33], this genetic association raises the possibility of TL1A having a positive role in protection from pathogens. Interestingly, studies in Vietnamese [120], Indian, and West African [121] leprosy cohorts have been unable to replicate genetic association with the TNFSF15 locus, suggesting population-specific effects or disease heterogeneity among the cohorts included [122].

Efforts to uncover the mechanism by which polymorphisms near TNFSF15 confer a risk for IBD have found that these single-nucleotide polymorphisms correlate with TL1A expression in monocytes [123], monocyte-derived macrophages [43], and T cells [124], as well as systemically in serum and in rectal mucosal biopsies [109]. These studies find increased TL1A expression with the IBD risk allele, although risk is defined as the opposite allele in the Ashkenazi Jewish study [123]. Recent expression quantitative trait loci studies, linking genetic to expression data across the whole genome, have also found these polymorphisms in the TNFSF15 locus associated with TL1A expression in monocytes [125] and influenza-infected dendritic cells [126]. However, the different ethnic populations and risk alleles examined, as well as the lack of concordance in cell types deemed responsible, leave unclear the downstream mechanism by which genetic control of TL1A expression alters the risk of developing IBD. Thus, although TL1A expression levels tend to associate with inflammation across a broad range of autoimmune and inflammatory diseases, this does not necessarily indicate a causal role at the genetic level. Genetic variants often exhibit cell-type-specific effects on gene expression, and it is therefore plausible that variants at the TNFSF15 locus affect gene expression in particular cell types that are only relevant in a subset of inflammatory diseases.

PROGRESS AND CHALLENGES IN UNDERSTANDING THE BIOLOGY OF TL1A

Since the initial discovery of TL1A as a T cell costimulator in 2002, great effort has gone into uncovering the biology of this TNF family ligand and its cognate receptor DR3. From initial research concentrating on its role in T cells, the field has grown to demonstrate its expression and function on a vast array of cell types and tissues (Fig. 1 and Tables 1 and 2). Multiple murine models have contributed to our current understanding of the role of TL1A as a T cell and ILC costimulator that is necessary for robust, local immunological responses and that participates in both autoimmunity and infection control (Table 3). The contribution of TL1A to human disease susceptibility has been highlighted by discoveries of disease-associated polymorphisms within the TNFSF15 locus. Furthermore, many additional autoreactive diseases have been associated with increased expression of TL1A both systemically and locally at the site of inflammation (Table 4). These findings suggest that, similar to the wide success of targeting TNF therapeutically [127], blockade of TL1A signaling might be beneficial in associated diseases. However, the role of TL1A in ongoing disease pathology with respect to cause and effect has yet to be determined and remains a key question that must be answered if TL1A-modulating therapeutics are to be considered.

AUTHORSHIP

A.C.R., J.R.F., F.M., E.T.H., O.G., and R.M.S. contributed to the development of ideas, literature research, writing, and editing the text and figures.

Glossary

CD

Crohn’s disease

DcR3

decoy receptor 3

DR3

death receptor 3

EAE

experimental autoimmune encephalomyelitis

FADD

Fas-associated death domain protein

GWAS

genome-wide association study

iT_reg

in vitro regulatory T cell

IBD

inflammatory bowel disease

ILC

innate lymphoid cell

PBC

primary biliary cirrhosis

RA

rheumatoid arthritis

TL1

TNF-like factor 1

TL1A

TNF-like factor 1A

TRADD

TNFR-associated death domain protein

TRAF

TNFR-associated factor

T_reg

regulatory T cell

UC

ulcerative colitis

DISCLOSURES

R.M.S. and F.M. hold intellectual property related to TL1A research.