Abstract
Objective. Tumor necrosis factor (TNF) receptor-associated periodic syndrome (TRAPS) is an autosomal-dominant multisystemic autoinflammatory condition. Patients display different mutations of the TNF receptor superfamily 1A gene (TNFRSF1A), coding for a nearly ubiquitous TNF receptor (TNFR1). No TNFRSF1A mutation has been identified in a proportion of patients with TRAPS-like phenotype.
Methods. We investigated mechanisms downregulating the TNF-induced inflammatory response such as (1) receptor shedding, producing a secreted form acting as a TNF inhibitor; (2) receptor internalization with subsequent induction of apoptosis; and (3) negative regulation of nuclear factor-κB (NF-κB) transcription. We analyzed the sequence of genes known to play a pivotal role in these pathways, in 5 patients with TRAPS symptoms and showing shedding and/or apoptosis defects, but without mutations of the TNFRSF1A gene.
Results. Sequence analysis of 3 genes involved in TNFR1 shedding (ERAP1, NUCB2, RBMX) and 3 genes involved in negative regulation of NF-κB signaling (TNFAIP3, CARP-2) or NF-κB transcription (ZFP36) revealed only a few unreported variants, apparently neutral.
Conclusion. Our study rules out any involvement in the pathogenesis of TRAPS of some of the genes known to regulate TNFR1 shedding and TNF-induced NF-κB signaling and transcription. Gene(s) responsible for TRAPS-like syndrome remain to be investigated among currently unidentified genes likely involved in these pathways, or by applying the genome-wide function-free sequencing approach.
- TUMOR NECROSIS FACTOR RECEPTOR-ASSOCIATED PERIODIC SYNDROME
- TNFR1 SHEDDING
- TNFAIP3
- CARP-2
- ZFP36
- CANDIDATE GENE SCREENING
Tumor necrosis factor (TNF) receptor-associated periodic syndrome (TRAPS) is an autosomal-dominant multisystemic autoinflammatory condition. Patients present with long-lasting recurrent fevers associated with abdominal pain, severe arthromyalgias, migratory rashes, fasciitis, periorbital edema, and frequent systemic AA amyloidosis as a longterm complication1. TRAPS has been associated with at least 64 different mutations of the TNF receptor superfamily 1A gene (TNFRSF1A; GenBank NM_001065.2) encoding for the transmembrane TNFR1 protein, also known as p55 TNFR (Institut de Genetique Humaine, CNRS-UPR1142, Montpellier, France; Website: http://fmf.igh.cnrs.fr/ISSAID/infevers). In particular, most TRAPS-associated TNFRSF1A mutations are missense substitutions mainly affecting the highly conserved cysteine residues in the extracellular cysteine-rich domains that are involved in the correct folding of the extracellular portion of the protein by forming disulfide bonds2. Indeed, TNFRSF1A mutations have been identified in a minority of patients and mainly in association with a positive family history3,4.
The precise pathogenetic mechanisms underlying TRAPS remain unresolved. TNFR1 is constitutively expressed as a trimeric protein on the cell membrane of most tissues. As well as TNFR1, TNF also binds another membrane receptor, TNFR2 (or p75 TNFR), that is typically found on cells of the immune system. The binding of TNF to TNFR1 results in immediate nuclear factor-κB (NF-κB) activation and subsequent apoptosis. Upon activation, cell-surface receptors are cleaved off by metalloproteinases, producing a secreted form that acts as a natural inhibitor of TNF activity. This process is known as “shedding” and this defect was originally proposed as a possible cause of the TRAPS phenotype2. Moreover, a constitutive release of full-length TNFR1 into the membranes of 20–50 nm exosome-like vesicles, capable of binding TNF, has been described in endothelial cells5. Finally, after binding TNF, the TNFR1 complex is also able to internalize into the cytoplasm, where the proapoptotic proteins FADD and caspase 8 are recruited. This pathway is induced when the NF-κB activity subsides, thus discontinuing the expression of antiapoptotic factors, and represents an additional means of downmodulating the TNF activity6. Recently, the generation of mouse models carrying Tnfr1 missense mutations homologous to that described in patients with TRAPS suggests that mutant Tnfr1 protein accumulates intracellulary, where it activates JNK and p38 signaling in a ligand-independent fashion. An anomalous response was also reported upon stimulation with low doses of lipopolysaccharide and other immune stimuli, which induced enhanced production of inflammatory cytokines and chemokines7.
On the basis of these considerations, we analyzed candidate genes possibly involved in the development of a TRAPS-like autoinflammatory phenotype in 5 patients with symptoms of this periodic fever, showing defects in the shedding and/or apoptosis, but without mutations, in the TNFRSF1A gene. In particular, we focused on (1) the 3 genes known to be involved in TNFR1 shedding, namely ERAP1 (endoplasmic reticulum aminopeptidase 1) and its partners nucleobindin 2 and RBMX (RNA-binding motif gene, X chromosome)8,9,10; (2) 2 genes responsible for negative regulation of NF-κB activation and TNF-induced apoptosis, namely TNFAIP3 (tumor necrosis factor alpha-induced protein 3) and CARP-211,12; (3) ZFP36, whose deficient mouse model develops overproduction of TNF-α13.
MATERIALS AND METHODS
Patients
Five patients with clinical phenotype and cellular defects resembling TRAPS, and showing no mutation in the whole coding portion and flanking intronic sequences of the TNFRSF1A gene, were selected. No other affected family member was reported for any of these patients. Our study protocol was approved by the Ethics Committee of the Gaslini Institute and informed consent obtained from all patients or their parents before enrolment into the study.
Mutation screening
Coding sequences of the following genes were analyzed by polymerase chain reaction (PCR) and direct sequencing: ERAP1, NUCB2, RMBX, TNFAIP3, CARP2, and ZFP36. PCR conditions and primer sequences are reported in Table 1. Primers were designed to include exons and exon-intron boundaries in the respective amplimers. PCR products were enzymatically purified using ExoSAP-IT (GE Healthcare) and directly sequenced using Big Dye v1.1 and a 3130 automated sequencer (Applied Biosystems, Foster City, CA, USA).
Polymerase chain reaction (PCR) conditions and primer sequences.
TNF-induced shedding and apoptosis on circulating monocytes
Monocytes from patients with TRAPS-like syndrome and healthy controls were obtained from blood samples, washed twice, resuspended in RPMI-1640 medium (Sigma, St. Louis, MO, USA) supplemented with penicillin/streptomycin, L-glutamine, nonessential amino acids (Bio-Whittaker, Walkersville, MD, USA), and 10% fetal calf serum (FCS, complete medium; Invitrogen, Carlsbad, CA, USA) at 1 × 106 cells/ml, and seeded in 12-well plates. Apoptosis was induced by treatment with 30 ng/ml recombinant TNF-α (PeproTech EC, London, UK) and 1 μg/ml cycloheximide (Sigma) for 4 h at 37°C, and assessed by staining with fluorescein-2-2 isothiocyanate-conjugated human annexin V (Bender MedSystems, Vienna, Austria) and propidium iodide. Shedding was induced by PMA treatment, and assessed after incubation with anti-p55 TNFR1-PE monoclonal antibody (R&D Systems, Minneapolis, MN, USA) for 30 min at 4°C. In every case, treatments were followed by cell washing and cytofluorimetric analysis (FACScan; BD Biosciences, San Jose, CA, USA), as reported14.
RESULTS
A total of 5 patients were selected among a large set of patients with autoinflammatory syndrome, with the following criteria: (1) presence of TRAPS-like clinical manifestations and/or positive diagnostic score indicative for a TRAPS phenotype15; (2) exclusion of severe, highly penetrant causative mutations of the coding sequence of the TNFRSF1A gene; and (3) evidence of a functional defect, assessed in patients’ monocytes, of either shedding or TNF-induced apoptosis14.
These patients displayed long-lasting episodes of fever associated with clinical manifestations typically observed in TRAPS (Table 2), as well as a functional defect (impaired shedding or TNF-induced apoptosis) associated with the TRAPS phenotype. In particular, in contrast to healthy controls, stimulation of patients’ monocytes with either TNF and cycloheximide or PMA promoted neither apoptosis nor shedding, respectively, with values within the range for TRAPS patients (Table 3). Significant differences of induced apoptosis and shedding between control and TRAPS cells had been assessed in a previous study14, and results of the current study clearly showed that patients had characteristics of TRAPS patients carrying TNFRSF1A mutations. Therefore all patients underwent molecular analysis for candidate genes possibly involved in the pathogenesis of the TRAPS syndrome phenotype.
Characteristics of patients with TRAPS-like syndrome: clinical features and functional cellular defects and genotypes at the TNFRSF1A and other candidate gene loci.
Functional findings in patients with TRAPS-like syndrome. Data are percentages.
DNA obtained from peripheral blood mononuclear cells of the 5 patients listed in Table 2 was analyzed under conditions reported in Table 1 for possible mutations of the coding sequence of the genes ERAP1, NUCB2, RMBX, TNFAIP3, CARP2, and ZFP36. As reported in Table 4, one of these patients was found to carry the missense variant p.R92Q, proposed to represent a low-penetrant hypomorphic mutation16. Nonetheless, his disease phenotype was quite severe, characterized by long-lasting fever episodes (> 15 days) with painful erythema of legs and arms and muscle-skeletal pain. An episode of fasciitis was also referred. For this reason he was included in our study.
TNFRSF1A and other candidate genes and variants detected in their coding sequences.
In Table 4 we give all the possible variants detected in the coding regions of these candidate genes. No variant was found in RMBX, TNFAIP3, and CARP2 genes. Among the variants identified in the ERAP1 gene, p.H417H displayed a frequency of the T allele significantly different in our patients compared to that recorded in HapMap (p = 0.0002), while p.S453S, p.K528R, and p.L848L presented with frequencies slightly but not significantly higher in our patients. As the frequency comparison for p.S453S approached significance (p = 0.08), to identify potential additional splicing sites or altered exonic enhancers, we investigated variants p.H417H and p.S453S using the Spliceport software (Website: http://spliceport.cs.umd.edu/) and Rescue ESE (Website: http://genes.mit.edu/burgelab/rescue-ese/). No effect, however, was suggested by these in silico analyses.
The in-frame 3 nucleotide deletion found in the nucleobindin 2 gene has been recorded in the SNPs database: (http://www.NCBI.NLM.NIH.gov/projects/SNP/), although not reported yet in population studies. Therefore, we sequenced DNA samples from 64 control individuals and calculated a frequency of 0.33 for the variant allele, which was not significantly different from that estimated in our patients.
An unreported synonymous nucleotide variant was also identified in the ZFP36 coding sequence in one of the 5 patients. In silico analysis using the software noted above revealed no formation of additive splice sites or exonic enhancers.
DISCUSSION
To determine the molecular basis sustaining pathogenesis in a small set of patients with TRAPS-like syndrome, we focused on a number of genes that, once mutated, may at some point dysregulate the TNFR1-mediated response to TNF, thus likely promoting a TRAPS-like autoinflammatory condition.
We started from analysis of genes coding for some membrane proteins that have recently been described in association with TNFR1, promoting TNFR1 release from human vascular endothelial cells. These are (1) ERAP1 (endoplasmic reticulum associated aminopeptidase 1), also known as ARTS-1 (aminopeptidase regulator of TNFR1 shedding), a type II integral membrane protein that binds full-length TNFR1 and regulates both the constitutive release of TNFR1 exosome-like vesicles and the proteolytic cleavage of soluble TNFR1 ectodomains8; (2) nucleobindin 2 (NUCB2), a calcium-dependent ERAP1 binding partner, which associates with cytoplasmic TNFR1 prior to its commitment to either release pathway10; and (3) RBMX (RNA-binding motif gene, X chromosome), which associates with ERAP1, regulating both constitutive and inducible pathways9. With the exception of the ERAP1 synonymous variant p.H417H, which was associated with patients reported here and which requires further investigation to exclude its role in the maturation of gene transcript, our study has ruled out that the 3 genes involved to date in TNFR1 shedding play a role in the TRAPS-like phenotype. This suggests that other genes are involved or that the observed shedding defect is only a secondary effect in this periodic fever.
However, shedding is not the only mechanism that down-modulates TNF-induced NF-κB activation. TNFAIP3 (tumor necrosis factor alpha-induced protein 3), also known as A20, is an early NF-κB-responsive gene encoding for a dual ubiquitin-editing enzyme involved in the negative feedback regulation of NF-κB signaling. Although aspects of TNFAIP3 biology are unclear, its role in the regulation of TNFR1 signaling by ubiquitination has been demonstrated11. In particular, TNFAIP3 interacts with a multiprotein complex where it has been shown to remove the ubiquitin chains from lysine 63 of RIP1, a modification necessary for the translocation of NF-κB into the nucleus. Subsequently, TNFAIP3 catalyzes the addition of ubiquitin chains to lysine 48 of RIP1, promoting its degradation by proteasome and termination of NF-κB activation11. In addition, a recent report demonstrated that CARP-2, a protein with ubiquitin ligase activity, is recruited to early TNFR1 complexes inside the endocytic vesicles, where it targets internalized RIP for proteasome-mediated degradation, thus limiting TNF-induced NF-κB activation12. In light of this, we included TNFAIP3 and CARP-2 in our list of candidate genes. A further control mechanism relies on transcriptional regulation of NF-κB activity. Among the proteins involved in this process, ZFP36, also called TTP, a zinc finger-containing protein that destabilizes mRNA by binding AU-rich elements in their 3’UTR, has been described as a negative regulator of gene expression17. Since mice deficient in ZFP36 develop a severe inflammatory syndrome characterized by overproduction of TNF-α, we included this gene in our study13. Our study has also ruled out causative mutations in the coding sequences of these genes, suggesting that other, unidentified genes are involved in these pathways, and novel candidate genes for TRAPS-like syndrome remain to be investigated.
We applied a candidate gene approach to a small set of patients, carefully selected for their clinical symptoms and for impairment of TNF-induced shedding and/or apoptosis, reflecting anomalous TNFR function. The inflammatory cell response is emerging as a broad and complex cellular mechanism, with many control steps and cross-talk pathways inducing opposite effects; we believe that identifying the causative gene(s) for TRAPS-like disorders will require a more straightforward and function-free approach than the strategies applied to date. In this respect, the emerging potential of the next generation of exome sequencing analysis provides a novel method for studies in this area of medical genetic research.
Acknowledgment
We are grateful to Loredana Velo for excellent secretarial assistance.
Footnotes
-
Supported by Italian Telethon (grant GGP07236) and the European Community Seventh Framework Programme (the EUROTRAPS project).
- Accepted for publication February 9, 2011.
