Drs. D'Osualdo and Ferlito contributed equally to this work.
Neutrophils from patients with TNFRSF1A mutations display resistance to tumor necrosis factor–induced apoptosis: Pathogenetic and clinical implications
Article first published online: 28 FEB 2006
Copyright © 2006 by the American College of Rheumatology
Arthritis & Rheumatism
Volume 54, Issue 3, pages 998–1008, March 2006
How to Cite
D'Osualdo, A., Ferlito, F., Prigione, I., Obici, L., Meini, A., Zulian, F., Pontillo, A., Corona, F., Barcellona, R., Duca, M. D., Santamaria, G., Traverso, F., Picco, P., Baldi, M., Plebani, A., Ravazzolo, R., Ceccherini, I., Martini, A. and Gattorno, M. (2006), Neutrophils from patients with TNFRSF1A mutations display resistance to tumor necrosis factor–induced apoptosis: Pathogenetic and clinical implications. Arthritis & Rheumatism, 54: 998–1008. doi: 10.1002/art.21657
- Issue published online: 28 FEB 2006
- Article first published online: 28 FEB 2006
- Manuscript Accepted: 1 DEC 2005
- Manuscript Received: 6 JUL 2005
- Italian Ministry of Health (Ricerca Corrente)
- Fondazione C. Golgi, Brescia
To explore tumor necrosis factor (TNF)–induced apoptosis in neutrophils from patients with TNF receptor–associated periodic syndrome (TRAPS) and to correlate the results with the different kinds of TNFRSF1A mutations.
Two hundred sixty-five patients with clinically suspected inherited autoinflammatory syndrome were screened for mutations of the TNFRSF1A gene. Neutrophils were isolated from heparinized blood by dextran sedimentation and incubated with and without cycloheximide (CHX) and TNFα. Cell apoptosis was assessed by human annexin V binding, and caspase 8 activation was assessed by flow cytometry.
Twenty-one patients were found to carry a variant of the TNFRSF1A gene: 13 patients had an R92Q substitution, and 8 patients presented other missense substitutions, 1 splicing mutation, and 1 in-frame interstitial deletion. Neutrophil stimulation with TNF and CHX was associated with induction of apoptosis in 12 normal controls and in 10 subjects with the R92Q mutation. Conversely, neutrophils from 8 TRAPS patients with mutations of cysteine or threonine residues or interstitial deletion did not show any induction of apoptosis after stimulation. The incidence of the R92Q mutation among patients with recurrent autoinflammatory syndromes was similar to that observed in the normal population.
Resistance to TNF-mediated apoptosis is a feature in TRAPS patients who have mutations of cysteine residues or interstitial deletion, and may play a pathogenic role. The R92Q mutation does not appear to be significantly associated with TRAPS.
Tumor necrosis factor receptor (TNFR)–associated periodic syndrome (TRAPS; MIM 142680) is an autosomal-dominant autoinflammatory disorder characterized by recurrent fever attacks lasting >1 week associated with abdominal pain, severe arthromyalgias, rash, and periorbital edema (1–3). Systemic AA amyloidosis represents the most serious long-term complication, with a reported incidence ranging from 14% to 25% (4, 5). TRAPS is associated with mutations of the TNFR superfamily 1A (TNFRSF1A) gene, located on chromosome 12p13 (6). The 55-kd protein product of the TNFRSF1A gene (TNFR p55) is expressed in a number of cell types and, together with TNFR p75, represents 1 of the 2 membrane receptors specifically bound by the proinflammatory cytokine TNF (7).
With the exception of 1 exon–intron junction mutation (already demonstrated to interfere with correct transcript splicing) and 1 single amino acid deletion in exon 3 (5, 8), all the TNFRSF1A mutations reported in TRAPS patients are missense substitutions mainly affecting the highly conserved cysteine residues in the extracellular cysteine-rich domains (CRDs) involved in disulfide bond formation and in the folding of the extracellular portion of TNFR p55 (5, 6). However, the role of some other TNFRSF1A mutations still needs to be defined. In particular, the R92Q mutation has been reported in 1% of healthy individuals, a finding that has suggested an incomplete penetrance (5). Moreover, it has been demonstrated that most patients carrying the R92Q mutation display a more heterogeneous clinical presentation with a milder disease course and lower incidence of amyloidosis (2, 5, 8).
Following cell activation, the extracellular portion of both TNFR p55 and TNFR p75 may undergo metalloproteinase-dependent cleavage (shedding) from the cell membrane (9). Shedding of free TNFRs from the membrane produces a pool of soluble receptors that may scavenge circulating TNF by competing with membrane-bound receptors. This latter phenomenon represents an important strategy for the regulation of the effect of circulating free TNF during acute inflammation. It has been suggested that some TNFRSF1A mutations may interfere with the process of shedding, leading to a lack of appropriate TNF inhibition and therefore to uncontrolled inflammation (6). However, TRAPS patients display a normal shedding of TNFR p75 (6) that, under normal conditions, represents the prevalent source of circulating soluble TNFRs (10). Moreover, recent observations have shown that some TNFRSF1A mutations are not associated with a defect of shedding of TNFR p55, suggesting that additional mechanisms could be related to the pathogenesis of the disease (5, 11, 12).
In contrast to the p75 receptor, TNFR p55 is also able to induce cell apoptosis, via activation of the caspases cascade (13, 14). In fact, after binding with TNF, the TNF–TNFR p55 complex is internalized in the cytoplasm and recruits the proapoptotic proteins FADD and caspase 8 (15). During cell activation, this intracellular signaling pathway is inhibited by the continuous expression of antiapoptotic factors produced through the activation of the NF-κB pathway. When the NF-κB activity subsides, the proapoptotic signals inducible by the intracellular TNFR p55 complex lead to cell death (14). Thus, TNFR p55 plays a unique role in the control of TNF-induced apoptosis of activated cells, representing a second strategy for the down-modulation of TNF activity during acute inflammation.
Therefore, it is conceivable that a dysfunction in the proapoptotic pathway of TNFR p55 might also account for the defective regulation of the inflammatory response observed in TRAPS. In the present study, we investigated TNFR p55–induced apoptosis as well as TNFR p55 shedding in TRAPS patients, and we correlated the results with the genetics and clinical phenotypes.
PATIENTS AND METHODS
In April 2002, the G. Gaslini Pediatric Institute in Genoa, Italy established a nationwide facility for the genetic diagnosis of autoinflammatory disorders associated with recurrent fever in childhood. All patients enrolled until March 2005 fulfilled the following criteria: 1) recurrent fever flares and 2) at least 2 of the following symptoms during attacks: lymphadenopathy, gastrointestinal involvement (vomiting, diarrhea, abdominal pain), skin lesions (erythematous rash, erysipelas-like lesions, etc.), and musculoskeletal manifestations.
Peripheral blood samples from 12 healthy volunteers and 4 patients with juvenile idiopathic arthritis (JIA) were used as controls for immunologic studies. Lymphocytes from a total of 200 unrelated Italian blood donors, collected at the G. Gaslini transfusion center, were used as a source of control DNA samples for mutation screening.
Our study protocol was approved by the ethics committee of the G. Gaslini Institute. Informed consent was obtained from all patients' parents and control individuals before study enrollment.
Molecular testing was performed on DNA extracted from peripheral blood lymphocytes by standard methods. The extracellular region of TNFR p55, spanning from exon 1 to exon 6 of the TNFRSF1A gene, was analyzed by means of denaturing high-performance liquid chromatography (dHPLC) as previously described (3), using 2 melting temperatures for each polymerase chain reaction (PCR) fragment to distinguish between heteroduplex and homoduplex molecules. WaveMaker 4.0 software (Transgenomic, San Jose, CA) was used to predict the appropriate linear acetonitrile gradient. Primers for PCR amplifications, amplicon lengths, Mg++ molarity, and dHPLC temperature conditions are listed in Table 1. All PCRs were performed in a total volume of 25 μl and run with an initial 12 minutes of denaturation at 95°C followed by 30 cycles at 95°C for 30 minutes, 55°C for 45 minutes, and 72°C for 45 minutes, with a final extension at 72°C for 7 minutes. Exons with altered elution profiles in dHPLC assay were directly sequenced with appropriate primers. Amino acid positions are defined relative to the signal peptide-cleavage site.
|Exon, forward and reverse primer sequence||First nucleotide of the forward primer*||Amplicon, bp||Mg++, mM||Denaturing HPLC temperature, °C|
To detect possible homozygous conditions for mild mutations (i.e., R92Q), for each patient and each exon under study amplification products were mixed with an equal amount of those obtained from a control individual known to carry no variant alleles at the TNFRSF1A locus, denatured, reannealed, and analyzed for heteroduplexes. Patients found to carry TNFRSF1A mutations were also analyzed for genetic defects in mevalonate kinase (MVK) and Mediterranean fever (MEFV) genes according to protocols already reported (16, 17).
Study of TNF-induced apoptosis on circulating neutrophils
Neutrophils were isolated from heparinized blood samples by dextran sedimentation and then purified on a Percoll (Amersham, Uppsala, Sweden) density gradient. Cells were collected, washed twice, and resuspended in RPMI 1640 (Sigma, St. Louis, MO) supplemented with penicillin/streptomycin, L-glutamine, nonessential amino acids (BioWhittaker, Walkersville, MD), and 10% fetal calf serum (FCS; Invitrogen, Carlsbad, CA) at 1 × 106/ml. Cells were seeded in 24-well plates and incubated with and without 1 μg/ml cycloheximide (CHX; Sigma) and 30 ng/ml recombinant TNFα (PeproTech EC, London, UK) for 6 hours. Notably, CHX acts as an inhibitor of cell transcription, blocking the induction of NF-κB–regulated survival genes (18). Cells were then washed twice with phosphate buffered saline (Sigma) containing 1% FCS (staining buffer) before being tested for apoptosis and caspase activity. Cell apoptosis was assessed by staining with fluorescein isothiocyanate–conjugated human annexin V (Bender MedSystems, Vienna, Austria) and cytofluorimetric analysis (FACScan; BD Biosciences, San Jose, CA) according to the manufacturer's protocol. Caspase 8 activation was analyzed by flow cytometry using the Apofluor Green Caspase Activity Assay (FAM-LETD-FMK reagents; Enzyme Systems Products, Livermore, CA) according to the manufacturer's protocol (19). Results were expressed as the percentage of positive cells. CellQuest software (BD Biosciences) was used for data analysis.
TNFR p55 expression and shedding in response to phorbol myristate acetate (PMA)
Heparinized peripheral blood samples from TRAPS patients and normal controls were collected. The plasma was removed by centrifugation at 2,500 revolutions per minute for 10 minutes. The remaining cells were washed with complete medium and resuspended to the original volume of blood collected.
One-milliliter aliquots of the plasma-free blood were incubated with or without PMA (Sigma) for 10 minutes at 37°C. The samples were then centrifuged at 2,500 rpm for 10 minutes, and the supernatant was removed and stored at –80°C. The remaining cells were washed twice and resuspended to 1 ml with staining buffer.
An aliquot of cell suspension (100 μl) was incubated with phycoerythrin (PE)–conjugated anti–TNFR p55 monoclonal antibody (R&D Systems, Minneapolis, MN) for 30 minutes at 4°C. Erythrocytes were then lysed with Becton Dickinson lysing solution (BD Biosciences), and after 2 washes with staining buffer, cells were analyzed by flow cytometry. An isotype-matched PE-conjugated mouse Ig (Caltag, Burlingame, CA) was included as control. The level of membrane expression of TNFR p55 on leukocytes was evaluated by setting the marker beyond the negative peak of isotype control. Results were expressed as the percentage of positive cells or as fluorescence intensity. The levels of soluble TNFR p55 in supernatants and plasma were analyzed with a Quantikine enzyme-linked immunosorbent assay (R&D Systems) according to the manufacturer's protocol.
The end point of the statistical analysis was the difference in the percentage change of annexin V binding and caspase 8 activation between unstimulated and stimulated cells (for apoptosis) and before versus after stimulation (for shedding). The significance of the differences was calculated using Wilcoxon's test for paired data. Heterogeneity among the various subgroups was evaluated by Kruskal-Wallis analysis of variance (ANOVA). The Mann-Whitney U test was used in a post hoc analysis of differences between subgroups. The incidence of the R92Q mutation of the TNFRSF1A gene was compared between the 265 patients with suspected inherited autoinflammatory syndromes and the 200 healthy controls, by 2-tailed chi-square test.
Clinical and genetic characterization of the patients.
Among the 265 patients with clinically suspected inherited autoinflammatory syndrome who were screened for mutations of the TNFRSF1A gene, 21 presented heterozygous mutations of the extracellular portion, from exon 1 to exon 6 (Tables 2 and 3). The R92Q substitution was the most common mutation (13 patients). Eight patients presented other mutations, including 6 patients with missense substitutions (among which there were 4 mutations affecting cysteine residues and the already known T50M substitution) as well as 1 patient with an already reported splicing mutation and 1 patient with a novel in-frame interstitial deletion. In total, 4 of 8 TNFRSF1A mutations found in these patients had never been described before (see families 1, 3, 6, and 8) (Table 2). The 21 affected carriers of a TNFRSF1A mutation were also analyzed for defects of the MEFV and MVK genes. Two of them presented MVK mutations in association with R92Q (see below), whereas no patient displayed associated MEFV mutations.
|1, proband||2||3||4||5||6, proband||7, proband||8, proband|
|Age at disease onset, years||1||<1||4||1||6||4||9||2||10||2||8||1|
|Duration of fever, days||15||15||12||15||15||30||60||15||10||4||4||3|
|Number of spikes per year||6||6||6||2||3||1||1||1||3||12||15||60|
|Mutation of the TNFRSF1A gene||C52Y||C43R||C55Y||C88Y||T50M||c.586–612del27||c.194–14 G>A||D12E|
|Age at disease onset, years||<1||<1||8||22||1||1||2||<1||4||<1||3||2||5|
|Duration of fever, days||3||5||15||7||3||3||3||3||4||3||3||3||3|
|Number of spikes per year||13||12||4||3||12||13||4||20||12||12||9||24||12|
|Mutation of the TNFRSF1A gene||R92Q||R92Q||R92Q||R92Q||R92Q||R92Q||R92Q||R92Q||R92Q||R92Q||R92Q||R92Q||R92Q|
In 17 families, both parents, and other family members with clinical manifestations possibly related to TRAPS, were also analyzed for TNFRSF1A mutations. No individual belonging to the remaining 4 families was available for further analyses.
Family history was negative in all patients carrying an R92Q mutation, whereas 4 patients presenting different TNFRSF1A mutations had family members with TRAPS-related clinical manifestations. Therefore, in order to focus on the clinical and immunologic features related to the different TNFRSF1A mutations, patients were divided into 2 subgroups: 1) patients carrying a TNFRSF1A mutation different from R92Q and 2) patients with the R92Q mutation. The clinical characteristics of both subgroups are described in Tables 2 and 3.
Patients carrying mutations of the cysteine residues (families 1–4) displayed a severe disease course with episodes characterized by prolonged fever (mean duration 23 days) and additional symptoms related to a wide range of different disease-related clinical manifestations (Table 2). The mean age at disease onset was 3.7 years (range <1–9 years).
Patient 5 displayed a T50M mutation and experienced episodic long-lasting attacks of fever from age 2 years. Her father presented clinical manifestations consistent with TRAPS and developed severe amyloidosis in adulthood. He died at age 48 years from acute pancreatitis (Table 2). Patient 6, originating from Mauritius, displayed an early onset of the disease that rapidly progressed toward amyloidosis with severe renal impairment requiring dialysis treatment from age 19 years. A heterozygous deletion of 27 nucleotides in exon 6 was found in this patient (Table 2). Parents of this patient were not available for genetic testing since he was adopted at age 3 years. Interestingly, he recently came into contact for the first time with a brother, a 35-year-old man still living in Mauritius who underwent renal transplantation 3 years ago for AA amyloidosis. Patient 7 presented an intronic mutation already reported to cause a 4–amino acid insertion between exon 2 and exon 3. He had short but frequent episodes of fever associated with severe clinical manifestations (Table 2). At age 1 year, patient 8 began to develop frequent attacks of fever lasting 3 days and not associated with any other relevant clinical manifestation (Table 2). Most of the patients required medium-to-high cumulative doses of steroids to control duration and intensity of the symptoms. No patient was receiving immunosuppressive or biologic treatment at the time of the study.
Patients with the R92Q mutation presented peculiar clinical features different from those reported in patients carrying other TNFRSF1A mutations (Table 3). Six of 13 patients had onset of their disease during the first year of life. The mean duration of the fever episodes was 4.4 days (range 3–15 days). A considerable proportion of patients experienced exudative or erythematous pharyngitis, aphthous stomatitis, and cervical lymphadenopathy (Table 3). Notably, 5 of 13 patients satisfied all clinical criteria for the periodic fever, aphthous stomatitis, pharyngitis, and adenitis (PFAPA) syndrome (20). Moreover, patients 9 and 10 (described in detail elsewhere ) also presented 2 mutations in compound heterozygous state of the MVK gene. In the majority of these “R92Q” patients, a single dose of steroid taken at the onset of the fever attacks was effective in the control of the symptoms and shortened the natural duration of the episodes. In all the families tested, at least 1 of the parents was found to carry the R92Q mutation without presenting, however, any relevant clinical manifestation. In conclusion, patients carrying the R92Q mutation displayed a milder disease course in terms of duration of flares, intensity of disease-associated symptoms, and responsiveness to steroid treatment.
Notably, the mutation analysis of 200 healthy donors revealed the presence of the R92Q mutation in 9 subjects (2.25%). This incidence was very similar to that detected for the same mutation among the whole set of 265 recruited patients with autoinflammatory syndrome (13 R92Q alleles, corresponding to an incidence of 2.45%) (P = 0.84).
Defect of TNF-induced apoptosis in TRAPS patients with cysteine or threonine mutations and the interstitial deletion.
To investigate whether different mutations of TNFRSF1A could be associated with a defect of TNF-induced apoptosis in circulating leukocytes, freshly isolated neutrophils from 8 patients with cysteine or threonine substitutions or interstitial deletion (families 2–4 and patients 5 and 6) (Table 2), from 5 symptomatic patients with the R92Q mutation (patients 11, 15, 16, 19, and 20) (Table 3), and from 5 asymptomatic subjects with the R92Q mutation were stimulated with TNF and CHX and compared with untreated cells (Figure 1). At the time of the study, all the patients were in an attack-free interval of their disease (absence of fever and of major clinical manifestations). Nevertheless, some of them displayed a slight elevation of systemic markers of inflammation. The mean erythrocyte sedimentation rate (ESR) was 20.3 mm/hour (range 7–40) in patients with R92Q mutations and 23.5 mm/hour (range 10–42) in patients with cysteine or threonine mutations or interstitial deletion.
In all healthy controls, treatment with TNF in combination with CHX was associated with the induction of apoptosis, as shown by the variable but constant increase of the cellular binding of annexin V found in treated versus untreated neutrophils (median percentage change of annexin V–positive cells with respect to untreated cells +366%, range +85% to +718%; P = 0.002 by Wilcoxon's rank test) (Figures 1A and C). The same effect was observed in patients with the R92Q mutation (+135%, range +33% to +170%; P = 0.04) and in their asymptomatic relatives carrying the same mutation (+125%, range +60% to +583%; P = 0.04) (Figures 1A and C). Conversely, neutrophils of TRAPS patients with either mutations in cysteine or threonine or interstitial deletion did not show any induction of apoptosis after stimulation with TNF and CHX compared with untreated cells (median percentage change 0%, range −25% to +50%; P = 0.7) (Figures 1A and C).
The heterogeneity test among the various subgroups evaluated was highly significant (P = 0.004 by Kruskal-Wallis ANOVA test). At post hoc analysis, TRAPS patients with cysteine or threonine mutations or gene deletion displayed a clear lower percentage of change compared with healthy individuals (P = 0.0006 by Mann-Whitney U test) and compared with patients carrying R92Q mutations (P = 0.003 by Mann-Whitney U test) (Figure 1C).
Notably, untreated neutrophils from patients with the R92Q mutation and from TRAPS patients with other mutations displayed a higher percentage of annexin V–positive cells than that observed in healthy controls (Figures 1A and C). The same finding was also observed in 4 patients affected with JIA who served as disease controls (Figure 1C). Thus, spontaneous induction of apoptosis in untreated neutrophils appears likely to be related to the underlying state of inflammation presented by patients with different inflammatory conditions. It is noteworthy that neutrophils from patient 9, affected with hyperimmunoglobulinemia D with periodic fever syndrome (HIDS) and carrying an R92Q mutation, showed normal TNF-induced apoptosis (median percentage change +125%; data not shown).
The concomitant evaluation of the expression of activated caspase 8 after stimulation with TNF and CHX was consistent with that observed for annexin V, with a lack of caspase 8 activation after TNF-induced proapoptotic stimulation in TRAPS patients carrying cysteine or threonine substitutions or the interstitial deletion (Figures 1B and D). In contrast, similar levels of TNF-induced caspase 8 activation could be demonstrated both in controls and in patients carrying the R92Q mutation (Figures 1B and D). Thus, our results show that circulating neutrophils from TRAPS patients, with the exception of those patients carrying the R92Q mutation, have a defect in TNF-induced apoptosis.
Effects of different TNFRSF1A mutations on TNFR p55 shedding.
It has been suggested that a defect of TNFR p55 shedding from the surface of activated circulating leukocytes plays a pivotal role in the pathogenesis of TRAPS (6). This hypothesis has recently been disputed by observations showing normal receptor shedding both in patients carrying the R92Q mutation (5) and in patients with different cysteine mutations (11).
Leukocytes from TRAPS patients with cysteine mutations showed a clear capacity for shedding of TNFR p55 after activation with PMA (median percentage change of TNFR p55 expression after stimulation −48.6%, range −31% to −53%) (Figures 2A and B). However, the percentage of reduction was significantly lower compared with that in normal controls (−78%, range −60% to −89%) (P = 0.006 by Mann-Whitney U test) (Figure 2B). Similarly, patients with the R92Q mutation showed a lower percentage of reduction compared with that in normal controls (−67%, range −55% to −67%) (P = 0.04), although, in contrast to patients with cysteine mutations, they displayed lower basal percentages of TNFR p55–positive cells in unstimulated leukocytes (Figure 2B). Notably, the same behavior was also observed in leukocytes isolated from JIA patients (data not shown). Similar results were obtained when the analysis was performed with gating on neutrophils and when mean fluorescence intensity was considered (data not shown).
Thereafter, these results were compared with the extent of soluble TNFR p55 protein levels both in plasma and in the supernatants collected from unstimulated and stimulated leukocytes (Figures 2C and D, respectively). TRAPS patients with cysteine mutations displayed a trend toward lower plasma levels of soluble TNFR p55 (P = 0.06) (Figure 2C) and a significantly lower increase of soluble TNFR p55 (median increase 377 pg/ml, range 73–570) in supernatants after stimulation with PMA compared with healthy controls (median increase 556 pg/ml, range 498–950) (P = 0.02) (Figure 2D).
Consistent with the results obtained by flow cytometry, patients with the R92Q mutation displayed higher plasma levels of TNFR p55 compared with the other 2 subgroups (Figure 2C). Similarly, 3 patients with the R92Q mutation displayed higher basal levels of soluble TNFR p55 in the supernatants from unstimulated leukocytes, possibly related to ongoing inflammation at the time of blood sampling. Notably, 2 of these 3 patients had concomitant moderate elevation of ESR. Taken together, our findings support the hypothesis of a partial defect of shedding in TRAPS patients with cysteine mutations and a substantially different behavior in patients carrying the R92Q mutation.
Patient 6 had a deletion of 27 nucleotides in exon 6, which leads to the absence of 9 amino acids (amino acids 167–175) at the protein level. This spacer region, between the CRD4 and the transmembrane domain, is of particular interest because it contains the sites of cleavage of the extracellular portion of TNFR p55 from the cellular membrane by metalloproteinases (21). By flow cytometry, the leukocytes from patient 6 showed a 50% reduction of TNFR p55 expression after PMA stimulation. Plasma levels of soluble TNFR p55 were almost 25 times higher than those in normal controls (9,643 pg/ml). This finding could be partially related to the presence of renal impairment in this patient, leading to a defect of clearance of soluble TNFR p55. However, it is noteworthy that similar very high concentrations were also observed in the supernatants of both unstimulated and stimulated leukocytes from the same individual (data not shown), leading to the hypothesis that an increased spontaneous TNFR p55 shedding from membrane surface could be associated with this deletion.
In the present study, we have found an impairment in TNF-mediated neutrophil apoptosis in TRAPS patients carrying mutations involving cysteine or threonine residues or gene deletion. Conversely, neutrophils from patients with the R92Q mutation had a response to proapoptotic TNF-mediated stimuli that did not differ from that observed in healthy individuals and in their asymptomatic relatives carrying the same mutation. Moreover, patients with the R92Q mutation displayed a much milder disease course, and the overall incidence of R92Q mutation among patients with recurrent autoinflammatory syndromes was found to be similar to that observed in the normal population.
In our population of 21 patients with a molecular diagnosis of TRAPS, missense substitutions of cysteine or threonine residues were associated with a more aggressive disease course (Table 2). Patients carrying mutations not affecting the cysteine residues, namely, the interstitial deletion c.586–612del27, the splicing defect c.194–14 G>A, and the missense substitution D12E, had milder disease with a shorter duration of fever spikes (Table 2). Extension of mutation screening to all the exons encoding the extracellular domain allowed us to detect a molecular defect which otherwise would have been overlooked. This is the first large interstitial deletion in the TNFRSF1A gene described so far (c.586–612del27), affecting the most proximal portion of the extracellular region, adjacent to the transmembrane domain, which, including metalloproteinase cleavage sites (21), might be involved in receptor shedding, interfere with the appropriate membrane localization or folding of the extracellular portion, or affect other receptor functions. The actual functional implication of this deletion is under investigation.
Thirteen patients, representing >50% of those carrying a defect of the TNFRSF1A gene, showed the R92Q substitution, a missense mutation already reported both in patients affected with TRAPS and in control individuals (5). The allele frequency found in our series of patients with autoinflammatory syndromes (13 of 530, or 2.45%) did not differ significantly from the allele frequency found in normal chromosomes (9 of 400, or 2.25%). This is not in accordance with a previous report showing a significant difference in R92Q allele frequency between 137 TRAPS patients of various ethnicities and a combined set of 766 controls (5), a discrepancy which might be reconciled by a few considerations, as described below.
In 11 of 13 cases, we could demonstrate that the mutation was inherited from 1 asymptomatic parent, thus suggesting, along with its presence in the normal control population, a reduced penetrance for this substitution. Moreover, patients carrying the R92Q mutation displayed a milder disease course in terms of duration of flares, intensity of disease-associated symptoms, and responsiveness to steroid treatment (2, 8), so it is unlikely that most of these patients would have been diagnosed as being affected with TRAPS. Indeed, many of them could be affected with PFAPA syndrome, a generally benign and self-limited condition of unknown origin that is relatively frequent during childhood (20, 22). In this light, the R92Q mutation can be regarded as a low-penetrance variant with a mild and broad contribution to autoinflammatory human disease, most likely depending on other modifying genes or environmental factors, as evidenced by the characterization of a large and heterogeneous set of patients affected with a wide range of symptoms.
Additional mutations were found in patients 9 and 10. In particular, besides the R92Q mutation of the TNFRSF1A gene, these individuals have been shown to be compound heterozygotes for mutations of the MVK gene, thus allowing a molecular diagnosis of HIDS and confirming the absence of TRAPS. The concomitant presence of TNFRSF1A mutations possibly present also in the normal population, like the R92Q and P46L mutations, has been described in patients with mevalonate kinase deficiencies and HIDS phenotypes (23, 24).
Since the description of the causative gene for TRAPS (6), the functional implications possibly related to mutations of the TNFRSF1A gene have been the subject of intensive investigation. The actual relevance of the defect of shedding of TNFR p55 from the leukocyte membrane after cell activation has been challenged by the observation of a large variability of this phenomenon in accordance with the different TNFRSF1A mutations analyzed (5, 8, 11, 25).
In the present study, TRAPS patients carrying cysteine mutations displayed a partial defect of shedding of TNFR p55 from leukocyte membrane surfaces and reduced expression of the soluble protein in the supernatants after PMA stimulation compared with normal healthy controls. However, none of the patients displayed a complete loss of the capacity to shed TNFR p55 into circulation. These findings raise the question of whether mechanisms other than the lack of inhibition of free TNF could be involved in the abnormal and sustained inflammatory response observed during disease flares.
As stated above, an intriguing functional aspect of TNFR p55 is related to its capacity to trigger either cell activation via the NF-κB pathway or cell apoptosis via induction of the caspase cascade (14). Notably, the proapoptotic signaling complex involving TRADD, FADD, and caspase 8 is temporally and spatially separated by the above-mentioned pathway and requires the internalization of TNFR p55 in the cytoplasm for its initiation (15). The specific inhibition of NF-κB activation or the blocking of new protein synthesis with various stimuli (including CHX) leads to the prompt activation of the proapoptotic pathways, due to the lack of regulation by NF-κB–dependent antiapoptotic proteins (26).
The hypothesis that mutations of TNFRSF1A could be related to an increased resistance to TNF-induced apoptosis has recently been raised in a study by Siebert and coworkers. They showed a significant decrease of cell death after TNF stimulation in skin fibroblasts derived from 1 patient with the C43S TNFRSF1A substitution (25). In the present study, resistance to TNF-induced apoptosis and to caspase 8 expression was demonstrated in circulating neutrophils, which represent a cell subset closely related to the systemic inflammation occurring during the disease flares observed in TRAPS patients. Notably, the same behavior was unequivocally and persistently observed in 8 consecutive TRAPS patients with severe mutations and an aggressive disease course, but not in symptomatic and asymptomatic subjects carrying the R92Q mutation.
The defect of TNF-induced apoptosis in neutrophils from TRAPS patients raises new and intriguing perspectives on the possible role of this phenomenon in the pathogenesis of the disease. On the one hand, it is possible that this observation might simply be related to a generic dysfunction in the signaling of mutated TNFR. On the other hand, it should be noted that, consistent with the fact that all TRAPS-related mutations are located in the extracellular portion of the protein, the intracellular signaling of the cytoplasmic death domain of TNFR p55 could not be affected. In line with this, HEK 293 cells transfected with mutant recombinant forms of TNFRSF1A with substitution in the CRD1 of the protein have been shown to retain a normal capacity for cytokine production and induction of apoptosis (27). It is therefore possible that in pathologic conditions, conformational changes of the extracellular portion of the receptor could interfere with its correct internalization and with the subsequent activation of the proapoptotic pathway (15). These alternative hypotheses are currently under investigation.
In conclusion, we have found that TRAPS patients with a severe disease course have a defect in TNF-mediated neutrophil apoptosis that may likely contribute to disease pathogenesis. This defect is not present in patients carrying the R92Q mutation; this finding, together with the observations that these latter patients displayed a much milder disease course and that the overall incidence of the R92Q mutation among patients with recurrent autoinflammatory syndromes is similar to that observed in the normal population, suggests that the R92Q mutation might have a broader influence on susceptibility to inflammation.
The authors would like to thank Dr. Vito Pistoia for his continuous encouragement and for the revision of the manuscript and Dr. Maria Pia Sormani for her invaluable help in the statistical analysis.
- 3the French Hereditary Recurrent Inflammatory Disorder Study Group. The enlarging clinical, genetic, and population spectrum of tumor necrosis factor receptor–associated periodic syndrome. Arthritis Rheum 2002; 46: 2181–8., , , , , , et al, and