|Home | About | Journals | Submit | Contact Us | Français|
The TNF superfamily member, LIGHT (TNFSF14) is a key cytokine that activates T cells and dendritic cells, and is implicated as a mediator of inflammatory, metabolic and malignant diseases. LIGHT engages the Lymphotoxin-β receptor (LTβR) and herpesvirus entry mediator (HVEM, TNFRSF14), but is competitively limited in activating these receptors by soluble decoy receptor-3 (DcR3, TNFRSF6B). Two variants in the human LIGHT alter the protein at E214K (rs344560) in the receptor-binding domain and S32L (rs2291667) in the cytosolic domain, however, the functional impact of these polymorphisms is unknown. A neutralizing antibody failed to bind the LIGHT-214K variant indicating this position as a part of the receptor-binding region. Relative to the predominant reference variant S32/E214, the other variants showed altered avidity with LTβR, and less with HVEM. Heterotrimers of the LIGHT variants decreased binding avidity to DcR3, and minimized the inhibitory effect of DcR3 towards LTβR-induced activation of NF-κB. In patients with immune-mediated inflammatory diseases, such as rheumatoid arthritis, DcR3 protein levels were significantly elevated.
Immunohistochemistry revealed synoviocytes as a significant source of DcR3 production, and DcR3 hyperexpression is controlled by post-transcriptional mechanisms. The increased potential for LTβR signaling, coupled with increased bioavailability due to lower DcR3 avidity, provides a mechanism of how polymorphic variants in LIGHT could contribute to the pathogenesis of inflammatory diseases.
The mechanisms involved in the development and pathogenesis of autoimmune diseases remain unclear due to the complexity of multiple contributing factors, including infection and genes involved in regulating immune responses. Genetic variations in multiple genes involved in antigen recognition and cosignaling pathways regulating T cells have emerged as contributing factors, and as potential therapeutic targets for treating autoimmune diseases. Cosignaling systems can either stimulate or inhibit the activation of T cells, and together aid in maintaining homeostasis of the immune system. Manipulation of cosignaling systems in animal models can alter the pathogenesis of autoimmune diseases, or enhance immune responses to tumors (1–4). However, cosignaling systems often have multiple components and form complicated networks that are inadequately defined in most disease processes, making the consequences of therapeutic intervention difficult to predict.
LIGHT, a member of the TNF superfamily of cytokines (TNFSF14; homologous to lymphotoxins, exhibits inducible expression, and competes with HSV glycoprotein D for HVEM, a receptor expressed by T lymphocytes), acts as a cosignaling system for T lymphocytes (5, 6). LIGHT is type 2 transmembrane glycoprotein with a short cytoplasmic tail at the N-terminus and a C-terminal ectodomain containing the canonical TNF homology domain, which trimerizes (7, 8). The trimeric structure of the TNF related ligands promotes the clustering of specific cell surface receptors that in turn initiate signaling. LIGHT activates two cellular receptors, the herpes virus entry mediator (HVEM, TNFRSF14) and the lymphotoxin-β receptor (LTβR) (7). LIGHT also engages decoy receptor-3 (DcR3), a soluble TNFSF receptor lacking transmembrane and signaling domains, that probably acts to limit bioavailability of LIGHT (9, 10). The LIGHT-HVEM interaction selectively activates NF-κB RelA (11) that initiates transcription of genes involved in cell survival and inflammation. In contrast, LTβR ligation induces both RelA and RelB forms of NF-κB (12) that in turn induce expression of genes involved in homeostasis, such as tissue organizing chemokines (e.g., CCL21, CXCL13) and intercellular adhesion molecules (e.g., ICAM-1). LIGHT also directly regulates an inhibitory cosignaling pathway formed by the interaction of HVEM with Ig superfamily members, BTLA (B and T lymphocyte attenuator) and CD160 (13, 14). Together, LIGHT and its paralogous ligands, TNF, LTα and LTβ, and the Ig members, BTLA and CD160 form a multipathway cosignaling circuit that regulates inflammation and homeostasis of the immune system (6, 15).
LIGHT has emerged as a potential therapeutic target in inflammatory, metabolic and malignant diseases (16). Enforced expression of LIGHT in T cells induces a profound inflammatory disease focused in the gut and reproductive organs (17, 18), and blockade of the LIGHT/LTαβ pathways attenuated experimental autoimmune diseases (19). LIGHT is elevated in serum from patients with RA (20, 21) and may also play a role in dyslipidemia (22) and hepatic regeneration (23). Interestingly, the LIGHT system is specifically targeted by herpesviruses as part of their strategies of entry and immune evasion (24). Envelope glycoprotein D of herpes simplex virus (HSV)-1 and 2 binds HVEM blocking LIGHT (7), and gD activates HVEM, inducing the NF-κB transcriptional complex (11), and human cytomegalovirus orf UL144 encodes a mimic of HVEM that binds BTLA, stimulating inhibitory signaling (25). Persistent, lifelong infections caused by viral pathogens, such as herpesviruses, are considered environmental risk factors that may precipitate autoimmune disease in a host with appropriate genetic-based risks (26–28). Direct viral targeting of the LIGHT-HVEM-BTLA system may provide strong selective pressures affecting the evolution of these molecules.
The human LIGHT gene maps to chromosome 19p13.3 in a segment paralogous to the highly polymorphic MHC immune response loci (29), and within the region linked to inflammatory bowel disease locus-6 (IBD6) (30). The human LIGHT gene contains several variations in exonic and intronic regions that are functionally undefined. Here, we investigate the functional effect of two nonsynonymous variants in the coding region of human LIGHT. We demonstrate that the E214K is positioned near the receptor-binding region altering avidity for LTβR, and the S32L located in the cytosolic domain impacts the stability of LIGHT. Heterozygous combinations decreased the binding avidity to DcR3 minimizing the inhibitory effect of DcR3 towards LTβR-induced activation of NF-κB. DcR3 was elevated in the synovial fluid in patients with RA. Together, the variants of LIGHT may increase bioavailability and signaling due to decreased regulation by DcR3 and enhanced avidity for LTβR. We suggest the polymorphisms of the LIGHT system may reflect the proverbial double-edged sword: an increased advantage for resistance to pathogens may promote susceptibility to autoimmune disease.
Antibodies used in this study included mouse anti-human LIGHT recombinant Fab'2 (clone Gem1.2)(29), goat anti-LTβR (31), mouse anti-LTβR (BD-A8 clone, a gift from J. L. Browning, Biogen Idec, Cambridge, MA), mouse anti-HVEM mAb (clone 122, BioLegend, San Diego, CA), rabbit-anti-RelA/p65 Ab (C-20), anti-RelB (C-19) and anti-TRAF3 Ab (H-122) (Santa Cruz Biotechnology, Santa Cruz, CA). Human anti-human LIGHT mAb (clones F23 and E1) were made against recombinant human LIGHT using KM mouse technology essentially as described (32). Purified Fc fusion proteins, human LTβR-Fc and HVEM-Fc were produced and purified as described (25), and DcR3-Fc was purchased from R & D Systems Inc. (Minneapolis, MN).
The cDNA containing the polymorphic forms of LIGHT were made with a QuikChange site-directed mutagenesis kit (Stratagene, San Diego, CA) and confirmed by DNA sequencing of the entire coding region. The retroviral vector, pMIG-GFP was used to introduce various forms of LIGHT into EL4 cells (25). Flow cytometry-based binding assays were carried out as described (18, 33) and the calculated avidity yield values for those ligands match with other immobilized ligand binding assays (ELISA and plasmon resonance). For saturation binding and competition inhibition assays, graded concentrations of recombinant proteins (LTβR-Fc, HVEM-Fc, or DcR3-Fc) were diluted in binding buffer (2% FBS in PBS, pH 7.4 with 0.02% NaN3) and incubated for 60 min at 4°C. Goat anti-human Fc fragment (IgG)-specific antibody conjugated with R-phycoerythrin was used for detecting the Fc fusion proteins. Specific mean fluorescence intensity was obtained by subtracting the background fluorescence staining of the nontransduced parental EL4 cells or isotype matched control antibody (negative controls) from the experimental group. The KD and IC50 values were determined by nonlinear regression analysis with PRISM (version 4, GraphPad, San Diego), or by the Lineweaver–Burk plot.
The expression, stability and turnover of LIGHT was assessed using flow cytometry and calculated with the following equations. M = ph(1/s)C (Equation 1), where M = MFI of membrane LIGHT staining by flow cytometry; p = the rate of protein synthesis; h = the half-life of LIGHT; s = the rate of LIGHT shedding; and C = a constant of the equation. MCHX = h(1/s)C (Equation 2), where MCHX = MFI of membrane LIGHT staining of cycloheximide-treated cells (2 µg/ml for 6 hours). The relative rate protein synthesis between the reference and variants of LIGHT was obtained by dividing equation 1 by equation 2, M/MCHX = [ph(1/s)C]/[h(1/s)C], thus M/MCHX = p and => M32S-214E/M32S-214E, CHX = p32S-214E; => M32S-214K/M23S-214K, CHX = p32S-214K; => M32L-214E/M23L-214E, CHX = p32L-214E
The effect of the polymorphism on the half-life of LIGHT and the rate of LIGHT shedding was also examined since MCHX = h(1/s)C => M32S-214E, CHX = h1(1/s1)C (Equation 3); => M32S-214K, CHX = h2(1/s2)C (Equation 4); => M32L-214E, CHX = h3(1/s3)C (Equation 5). When equation 3 divided by equation 3, M32S-214E, CHX/ M32S-214e, CHX = [h1(1/s1)C]/[ h1(1/s1)C] = 1. When equation 4 divided by equation 3, M32S-214K, CHX/ M32s-214e, CHX = [h2(1/s2)C]/[h1(1/s1)C]. When equation 5 divided by equation 3, M32L-214E, CHX/M32s-214E, CHX = [h3(1/s3)C]/[h1(1/s1)C].
Example 1 (with equation 4 divided by equation 3), M32L-214E, CHX/ M32S-214E, CHX = [h2(1/s2)C]/[h1(1/s1)C] = h2s1/h1s2 = 1. When h1s2/s1h2 = 1 => h1s2 = h2 s1 => h1/h2 = s1/s2, if h1 = h2, s1 = s2, if s1 = s2, h1 = h2.
Example 2 (with equation 4 divided by equation 3), M32L-214E, CHX/M32S-214E, CHX = [h2(1/s2)C]/[h1(1/s1)C] = h2s1/h1s2 > 1. When h2s1/h1s2 > 1=> h2s1 > h1 s2, if s1 = s2, h2 > h1, if h2 = h1, s1 > s2.
Example 3 (with equation 4 divided by equation 3), M32L-214E, CHX/M32S-214E, CHX = [h2(1/s2)C]/[h1(1/s1)C] = h2s1/h1s2 < 1. When h2s1/h1s2 < 1=>h2s1 < h1 s2, if s1 = s2, h2 < h1, if h2 = h1, s1 < s2.
HeLa cells were cotransfected with the dual-luciferase reporter plasmids (pNFκB, Stratagene, San Diego, CA; pRL-TK, Promega, Medison, WI), and various LIGHT isoforms expressing EL4 cells were added to cell cultures overnight. Luciferase reporter assay with the Dual-Luciferase Reporter Assay System™ (Promega) was carried out as previously described (11). For the functional competition assays, graded concentrations of DcR3-Fc were added to the cell culture, and incubated overnight. Cell lysates were prepared and the luciferase activity was measured.
Levels of DcR3 in serum and synovial fluid were detected with the human DcR3/TNFRSF6B DuoSet ELISA Development kit (R&D Systems, Minneapolis, MN, U.S.A.) following the manufacturer’s instructions. Serum and synovial fluid samples were serially diluted for use in the ELISA assay.
Sections were cut from formalin-fixed, paraffin-embedded tissue samples and paraffin was removed. Heat-induced antigen recovery was performed using the Antigen Retrieval Reagent – Basic buffer (R&D Systems) followed by blocking of endogenous peroxidase, serum and biotin/avidin (as supplied in the anti-goat Cell & Tissue Staining Kit; R&D Systems). A goat anti-human DcR3/TNFRSF6B affinity purified polyclonal antibody was used as the primary antibody at 10 µg/mL and goat polyclonal IgG was used as a negative control. Detection was performed with the anti-goat Cell & Tissue Staining Kit, either using DAB (brown). Sections were counterstained with hematoxylin and mounted for microscopic analysis.
Serum and synovial fluid, or synovial tissue samples were obtained on the occasion of diagnostic or therapeutic procedures from patients suffering from rheumatoid arthritis (American College of Rheumatology criteria), spondylarthritis (European Spondyloarthropathies Study Group criteria), systemic lupus erythematosus or from osteoarthritis (American College of Rheumatology criteria). All patients were of Caucasian origin. Synovial tissue biopsies were gathered at the occasion of a needle arthroscopy as previously described, frozen and stored at −80°C (34). All patients provided written informed consent and the study was approved by the ethical committee of Ghent University Hospital.
Total RNA was prepared from synovial biopsies using the RNeasy kit and on-column RNase-Free DNase Set (Qiagen, Valencia, CA). RNA yield was estimated by spectrophotometric analysis and approximately 1µg was used per reverse transcription PCR reaction. Primers for detection of DcR3 and control mRNA by PCR were designed as described (35). Detection was performed with the Applied Biosystems 7000 Real-Time PCR System, using SYBR green (Invitrogen) fluorescence as the read-out. Relative expression levels were calculated upon normalization relative to the geometric mean of three housekeeping gene Ct values: cyclophylin A, hypoxanthine phosphoribosyl-transferase and glyceraldehyde-3-phosphate dehydrogenase.
Two validated, nonsynonymous, single nucleotide polymorphic variants have been identified within the coding region of LIGHT. The exonic variants reside at position E241K (rs344560) in the ectodomain and S32L (rs2291667) in the N-terminal cytosolic domain (Figure 1A). The E214K variation resulting from a G/A substitution is common in the US population (Figure 1B–D), with 214E variant the predominant allele (heterozygosity = 0.11). Significant differences in the distribution of variants were observed among prevalent ethnic groups (Table 1). In contrast, the 32L variant generated by a T/C substitution is relatively rare (heterozygosity = 0.006). S32 is one of three serines in the cytosolic domain of human LIGHT; the other two are located at position 9 and 27 (Figure 1A). Bioinformatic analyses indicated S9, S27, and S32 fit the consensus sequence for phosphorylation by cyclic AMP-dependent protein kinase A.
We assessed the effect of the cytosolic domain serine residues on the cell surface expression of LIGHT by flow cytometric analysis (Figure 2A). LIGHT expression was detected in EL4 cells expressing LIGHT32S-214E, 32S-214K, 32L-214E or 32L-214K as detected with the anti-LIGHT (Gem 1.2 Ab). GFP served as an internal control to normalize expression. Interestingly, the amino acid substitution of 32L resulted in a significant reduction of membrane bound LIGHT (Figure 2A).
To determine the basis of lower expression of the LIGHT 32L variant we quantified LIGHT protein expression and half-life. The MFI value obtained from membrane LIGHT staining (M) is directly related to the rate of LIGHT protein synthesis (p) and the half-life of LIGHT (h), and inversely related to the rate of membrane LIGHT shedding (s) as defined by M = ph(1/s) C, where C is a constant of the equation. Cycloheximide was used to block protein synthesis in EL4-LIGHT cells to measure protein stability. The MFI of membrane LIGHT staining was measured in EL4-LIGHT cells treated with cycloheximide (MCHX), and the relative rate of LIGHT protein synthesis between the reference LIGHT, LIGHT32S-214E and the variants were calculated as a ratio of M to MCHX (Figure 2B, left panel; see details of mathematical equations in the Materials and Methods section). The data demonstrated that the amino acid substitutions of S32L or E214K did not alter the rate of LIGHT synthesis (Figure 2B, left panel), suggesting perhaps the half-life or shedding of LIGHT may be altered by the L32 substitution. The MCHX of the LIGHT variants were compared with the MCHX of the reference LIGHT32S-214E. The 214K variant did not affect either the half-life or shedding of LIGHT, as the ratio between M32S-214K, CHX and M32S-214E, CHX was equal to 1. However, the ratio between M32L-214E, CHX and M32L-214K, CHX was 0.7 (significantly less than 1), suggesting 32L might result in a decrease of the half-life of LIGHT and/or an increase of LIGHT shedding. Western blotting to assess shed LIGHT revealed a reduced level from cells expressing the LIGHT32L variant relative to the reference LIGHT32S (Figure 2C). This observation was consistent with the level of LIGHT32L present on the cell surface with reference to LIGHT32S as well as LIGHT214K (Figure 2A), indicating that the 32L variant might not have a significant impact on the shedding of LIGHT in EL4 cells. To examine the possibility that the reduced level of membrane LIGHT32L expression was due to an intrinsic defect in membrane translocation, combinations of intracellular and membrane staining were performed. The S32L variant showed a significant reduction (~30%) of expression in both intracellular and membrane compartments (Figure 2D). Taken together, our results indicated that LIGHT-32L substitution appears to have a negative impact on the half-life or stability of the LIGHT polypeptide.
A molecular model of LIGHT (8) indicated the E214K residue is surface exposed and located in the G β-strand that lies adjacent to the anti-parallel E β-strand near the D–E loop (Figure 1B). Mutations in the D–E loop of LIGHT impact receptor binding (8) raising the possibility that the E214K polymorphism might affect receptor binding. To assess whether these nonsynonomous variants affected receptor binding, the LIGHT variants were assessed for binding with its cognate receptors using Fc fusion proteins of the ectodomains of LTβR, HVEM and DcR3 as surrogate receptors. Flow cytometric analysis demonstrated that the amino acid substitution of S32L or E214K did not prevent LIGHT from binding to its receptors (Figure 3A).
Two human monoclonal antibodies were assessed for binding these polymorphic variants of LIGHT. The F23 mAb bound both LIGHT variants with nearly equal avidity (EC50 = 0.08–0.1 µg/ml for 32S-214E, Figure 2B; 32L–214E, Figure 2C; and 0.1µg/ml for 32S-214K, Figure 2D). The S32L variants had no significant impact on the binding of either mAb. However, the binding of the E1 mAb to 214K variant was dramatically reduced compared to the 214E reference form (214K, EC50 >10µg/ml; 214E, EC50 = 0.5 µg/mlEC50l; Figures 2B–D). This result indicated 214E may be a critical part of the epitope recognized by the E1 mAb. Importantly, both of the anti-LIGHT mAb inhibited the binding of LTβR-Fc and HVEM-Fc to LIGHT 214E (Figure 2E), indicating the epitope containing the amino acid residue at position 214 topographically overlaps with the receptor-binding site.
A flow cytometric-based, saturation binding assay was used to measure binding avidity of the LIGHT variants to LTβR-Fc, HVEM-Fc and DcR3-Fc. LIGHT32S-214E expressing EL4 cells bound LTβR with an EC50 value of 3.04 µg/ml (Figure 4A). The polymorphic LIGHT32S-214K (Figure 4B, left panel) and LIGHT32L-214E (Figure 4B, right panel) variants showed a two-fold increase of binding avidity to LTβR-Fc compared to reference LIGHT32S-214E. Interestingly, the LIGHT variants (LIGHT32S-214K or LIGHT32L-214E) showed a modest, but significant difference in the binding avidities to HVEM and DcR3, but a two-fold increase in the binding avidity to LTβR-Fc with reference to the predominant form (Figure 4C).
Heteromeric combinations of LIGHT variants were coexpressed in EL4 cells including 32S-214E with 32S-214K, 32S-214E with 32L-214E, and 32L-214E with 32S-214K. We assumed that the two variants would show no bias in assembling into trimers. The combination of polymorphic LIGHT32L-214E with 32S-214K showed a two fold enhancement of binding to LTβR-Fc, but intriguingly, a three fold reduction in binding avidity to DcR3-Fc relative to the predominant form of human LIGHT (Figure 4D). This result indicated polymorphic heterotrimers influenced receptor binding. The increased avidity for LTβR and decreased avidity for DcR3 implicated these polymorphisms may alter both LTβR signaling and the bioavailability of DcR3, perhaps having a cooperative or synergistic impact on sustaining signaling by LIGHT-LTβR pathway.
LIGHT-induced LTβR signaling activates both RelA and RelB NF-κB pathways (12), while HVEM signaling activates only the RelA NF-κB pathway (11). HeLa cells express endogenous LTβR, but not HVEM (Figure 5A), and thus this cell line is useful in assessing the signaling function of LIGHT variants exclusively through LTβR-dependent NF-κB pathways. LIGHT32S-214K expressing EL4 cells were cocultured with HeLa cells at a 1:1 ratio overnight and nuclear translocation of RelA and RelB in HeLa cells was assessed by Western blot of nuclear extracts. NF-κB RelA (Figure 5B, upper panel) and RelB (Figure 5B, bottom panel) were translocated specifically when cocultured with EL4 cells expressing either LIGHT32S-214K or LIGHT32S-214E, but not EL4 cells with empty vector. In addition, HeLa cells transfected with an NF-κB-dependent luciferase reporter was used to assess the capacity of LIGHT32S-214K variant in mediating LTβR dependent NF-κB signaling (36). The results showed that both allelic forms of LIGHT induced a similar level of NF-κB dependent reporter gene expression (Figure 5C).
Heterozygous combinations of LIGHT subunits (32L-214E and 32S-214K) expressed in EL4 cells exhibited higher avidity binding of LIGHT to LTβR, but simultaneously a reduced affinity for DcR3 relative to the reference form of LIGHT (as indicated in Figure 4D). The interaction between the heterotrimeric form of LIGHT32L-214E/32S-214K and LTβR in the presence of DcR3-Fc was further evaluated with a flow cytometric-based competition assay (Figure 6A, left panel). The IC50 for DcR3-Fc to compete with the binding of LTβR to a heterotrimeric LIGHT32L-214E/32S-214K was ~2.5 times higher than the reference form of LIGHT32S-214E, and two fold higher for the 32S-214K form of LIGHT, respectively (Figure 6A, right panel). The results indicated that 2.5 fold more DcR3-Fc was needed to inhibit binding of LTβR-Fc to heterotrimeric LIGHT32L-214E/32S-214K than to the reference form of LIGHT.
It appeared that the combination of LIGHT32L-214E and 32S-214K could potentially be a stronger activating signal for LTβR pathway. The NF-κB reporter luciferase assay in HeLa cells was used to assess the efficacy of LIGHT32L-214E/32S-214K in activating LTβR signaling. The LIGHT32L-214E/32S-214K expressing EL4 cells did not further enhance NF-κB activity compare to the predominant form of LIGHT expressing EL4 cells (EL4-LIGHT32S-214E) (Figure 6B). This luciferase assay using EL4:HeLa cell ratio (1:1) is in the linear range and responsive to changes in ligand concentration (36).
To determine if the altered avidity of the LIGHT32L-214E/32S-214K for DcR3-Fc impacted the ability of DcR3 to antagonize LTβR mediated signaling, HeLa cells were cocultured with EL4-LIGHT32S-214E or EL4-LIGHT32L-214E/32S-214K in the presence of graded concentrations of DcR3-Fc. The results showed that the combination of LIGHT variants, 32L-214E and 32S-214K, minimized the inhibitory effect of DcR3 towards LTβR signaling induced by LIGHT (Figure 6C). Both the in vitro competition binding (Figure 6A) and functional blocking assays (Figure 6C) demonstrated the higher potential of heterozygous 32L-214E/32S-214K form of LIGHT to activate LTβR compared to the reference form of LIGHT in the context of DcR3. Together, the increased potential for LTβR signaling, coupled with increased bioavailability of both LIGHT and DcR3 due to lower DcR3 affinity suggests a plausible mechanism that polymorphic variants in LIGHT could contribute to the pathogenesis of inflammatory-mediated disorders.
The contribution of LIGHT polymorphisms to DcR3 binding and the potential for increased signaling stimulated an examination of DcR3 in inflammatory conditions. We assayed the expression of DcR3 protein in synovial fluid and plasma obtained from patients with RA (n = 31), spondylarthritis (SpA) (n = 43) and osteoarthritis (OA) (n = 30) using ELISA (Figure 7A). All RA, OA and SpA patients (except for two SpA) had detectable DcR3 protein in synovial fluid (Figure 7A). However, there was a remarkable increase of DcR3 in the RA cohort (mean ~40 ng/ml) as compared to either the SpA or OA groups. Although previous work had shown elevated DcR3 levels in RA synovium, these results indicate that local production of this molecule is significantly more pronounced in RA patients in comparison with OA and SpA, which are distinct arthritic conditions.
DcR3 levels in serum of RA (N = 64), SpA (N = 78) and OA (N = 43) patients were markedly lower (e.g., ~2 ng/ml in RA) compared to the synovial fluid (Figure 7B). Only a minority of SpA (n = 18) and OA (n = 9) patients had detectable levels of circulating DcR3, whereas those in healthy individuals were always low. In contrast, a statistically significant increase was observed in the RA group when compared to either the SpA or OA patient groups. There appeared to be no significant correlation between local DcR3 levels in the synovial fluid and those detected in paired serum samples (data not shown). Taken together, these data indicate that DcR3 is expressed in both local and peripheral compartments and significantly increased in RA patients. In contrast to RA patients, a subset patients (22/84) with systemic lupus erythematosus (SLE) expressed detectable levels of DcR3 in serum (Figure 7C), although these patients showed significantly higher levels of protein when compared to the normal control group (n = 20).
The observations that RA synovial fluid contained relatively high levels of DcR3 protein prompted an evaluation of its expression in tissues by immunohistochemistry (Figure 7D). DcR3 was readily detected in synovial biopsies in the majority of RA patients and a few of the OA patients, which in line with the results obtained from the synovial fluid. However, immunohistochemistry also demonstrated substantial amounts of DcR3 protein in SpA patients. Blindly scored sections analyzed for DcR3 content revealed a difference between RA and OA samples, whereas levels in SpA synovial tissue were only slightly lower compared to RA samples (Figure 7E). The highest levels in both conditions were found in the synovial lining and sublining regions, in sharp contrast with the absence in the synovial endothelium. These results confirm the observation that synovial tissue is a significant source of local DcR3 production under certain arthritogenic conditions. Interestingly, the mRNA expression levels in total RNA purified from RA, SpA and OA synovial biopsies showed no detectable differences between the various conditions (Figure 7F). The molecular mechanism that underlining the evaluation of DcR3 under arthritogenic conditions remains to be defined. The increase in the bioavailability of DcR3 in the synovial fluid appears to be regulated by post-translational mechanisms, which may involve protein processing and secretion as well as the availability of membrane-anchored DcR3 binding molecules, such as LIGHT or heparin (37).
We demonstrate that the signaling potential of LIGHT is affected by two polymorphisms at positions E214K and S32L. The 214 position in LIGHT directly influenced the binding avidity to a neutralizing human antibody and the LTβR, indicating this residue is located in or near the receptor-binding region. This result provides an important consideration for the use of inhibitors of LIGHT, such as decoy receptors or antibody, in analyzing patient responses in clinical trials. The 32L variant located in the intracellular domain lowered the avidity of binding to DcR3 and decreased the membrane expression of LIGHT. Increased bioavailability of LIGHT due to lower avidity for DcR3, and enhanced binding avidity of LIGHT for LTβR could combine to increase signaling activity and predispose to unwarranted inflammation.
The transcriptional regulation of LIGHT, like TNF, may also contribute to its bioavailability. Several polymorphisms located in the 5' promoter region impact the transcriptional activity of LIGHT (38, 39). Studies in mice indicated that sustained expression of LIGHT in T cells caused significant pathology (17, 18). Interestingly, the promoter haplotype polymorphism reported by Kong et al. (39) showed the haplotype with low expression was associated with vascular dementia. Although the opposite of what might have been predicted, this result implies a role for shedding and soluble LIGHT, perhaps acting through HVEM-BTLA pathway to protect the brain vasculature. This notion is supported by the experiments that demonstrated the HVEM-BTLA pathway protects the intestinal mucosa from immune damage (40).
LIGHT is also controlled by cleavage from the cell surface and binding to soluble decoy receptors. LIGHT is shed from the cell surface by proteolysis creating a soluble form that can bind all three of its receptors (29). Our data showed that EL4 cells expressing the LIGHT32L variant shed a reduced level of soluble LIGHT relative to the reference 32S. Although it is reasonable to deduce that the S32L substitution might have an negative impact on the shedding of LIGHT, the reduction of shed LIGHT from the 32L variant may well reflect on the amount of LIGHT present on the cell surface.
In addition, the LIGHT variant 32L-214E and 32S-214K in heteromeric complexes decreased the avidity for DcR3 and limited the capacity of DcR3 to inhibit activation of the LTβR by cell associated LIGHT. The mechanism of how this combination of LIGHT variants in the heterotrimers altered DcR3 avidity is not clear. It could be a direct effect on binding, or an indirect effect, perhaps related to the stability of LIGHT trimer, or modification of this putative phosphorylation site provides another intriguing possibility. Studies in progress are addressing this issue. However, the distinctive binding characteristics of the LIGHT variants in relation to DcR3 and LTβR are likely to have an impact on the bioavailability of both LIGHT and DcR3 in the tissue microenvironment, which may influence immune activation and inflammation.
Patients with active rheumatoid arthritis (21) have elevated serum levels of soluble LIGHT, implicating the potential contribution of LIGHT to the autoimmune disease process. Concurrently, elevated levels of DcR3 were identified in the synovial fluid and serum, and in tissues with active disease from patients with RA and SLE, but not osteoarthritis, implicating the DcR3 is a component of the inflammatory response mechanism. Our results are supported by studies from Bamias et. al., (41) and Hayashi et. al., (42) that showed increased DcR3 in RA and other reports of elevated DcR3 in SLE (43, 44). In RA patients, the constitutive expression of DcR3 mRNA was found to be at the similar level in normal and disease tissue, indicating that expression of DcR3 protein is probably under dynamic control by post-translational processes. Moreover, the bioavailability of free DcR3 in the synovial fluid and serum is likely to be influenced by the presence of other DcR3 binding molecules, such as TL1A, FasL, as well as heparin.
Our data showed that the LIGHT polymorphisms altered binding avidity to DcR3. Although an elevated level of DcR3 was strongly associated with active RA, the physiological role of DcR3 in the pathogenesis of autoimmune RA remains to be further defined. We suggest the possibility that DcR3 could play an enhancing role in autoimmune-mediated responses by blocking soluble LIGHT binding the HVEM-BTLA cis complex as an inhibitory signaling mechanism (36). This hypothesis is supported by the development of an autoimmune-like syndrome in mice expressing human DcR3 transgene, which lack the DcR3 gene, potentially revealing a role for LIGHT, TL1A or FasL (45).
In addition, LIGHT, LTβR and DcR3 may play a role in liver inflammation (17, 18) and the development of hepatocellular carcinoma (HCC) (46, 47). Chronic liver inflammation associated with Hepatitis B and C virus infections may be a contributing factor for the development of HCC. Two recent studies showed elevated DcR3 levels in the clinical isolated HCC samples (48, 49). Furthermore, Haybaeck et al. recently demonstrated a functional link between LTβR signaling and the formation HCC (50). Intriguingly, relative to Caucasians, there is a significantly higher incidence liver cancer in the African American and Asian groups (51), which also have higher allelic frequencies of the LIGHT-214K polymorphism. Our current findings, in particular, the distinctive binding features of the LIGHT variants in relation to DcR3 and LTβR, suggest the LIGHT polymorphisms could play a pathogenic role in the development and progression of liver inflammation and the formation of hepatocellular carcinoma.
Our results provide mechanistic insight into the effect of the polymorphic variants on signaling and bioavailability of LIGHT and DcR3. These LIGHT variants, acting individually or in concert as a haplotype, can alter the signaling potential of LIGHT. Variants in other components of the LIGHT cosignaling circuit, such as HVEM, BTLA, CD160 or LTβR might create haplotypes with disease linkages, although at this time such haplotypes are unknown. Pathogens place powerful selective pressures on genes involved in controlling immune responses. The evidence that α- and β-herpesvirus specifically target pathways involving LIGHT (7, 11, 25, 36) provides an argument for the natural selection of variants that might be more effective in controlling virus infections. A consequence of this natural selection of enhanced signaling may predispose towards the development of autoimmune disease and/or cancer, the cut of the proverbial double-edged sword.
The authors wish to thank the assistance of M. Macauley, B. Ware, B. Woods and N. Huang for technical assistance, and C. Benedict for comments.
1This research was supported by grants from the US Public Health Service, National Institutes of Health (CA069381, R37AI33068, and AI048073 to CFW), the Uehara Memorial Foundation (HS), and Fund for Scientific Research Flanders and a concerted action grant of Ghent University (DE).