Cell, Vol. 97, 133–144, April 2, 1999, Copyright 1999 by Cell Press

Germline Mutations in the Extracellular Domainsof the 55 kDa TNF Receptor, TNFR1, Define a Familyof Dominantly Inherited Autoinflammatory Syndromes

Summary

Autosomal dominant periodic fever syndromes arecharacterized by unexplained episodes of fever andsevere localized inflammation. In seven affected fami-

Michael F. McDermott,1,13,14 Ivona Aksentijevich,2,13

Jerome Galon,3,13 Elizabeth M. McDermott,5

B. William Ogunkolade,1 Michael Centola,2

Elizabeth Mansfield,2 Massimo Gadina,3

Leena Karenko,6 Tom Pettersson,6

lies, we found six different missense mutations of theJohn McCarthy,7 David M. Frucht,3

55 kDa tumor necrosis factor receptor (TNFR1), five ofMartin Aringer,3 Yelizaveta Torosyan,2

which disrupt conserved extracellular disulfide bonds.Anna-Maija Teppo,6 Meredith Wilson,8

Soluble plasma TNFR1 levels in patients were approxi-H. Mehmet Karaarslan,9 Ying Wan,10

mately half normal. Leukocytes bearing a C52F muta-Ian Todd,5 Geryl Wood,2 Ryan Schlimgen,4

tion showed increased membrane TNFR1 and reducedThisum R. Kumarajeewa,1 Sheldon M. Cooper,9

receptor cleavage following stimulation. We proposeJohn P. Vella,11 Christopher I. Amos,10 John Mulley,12

that the autoinflammatory phenotype results fromKathleen A. Quane,7 Michael G. Molloy,7

impaired downregulation of membrane TNFR1 andAnnamari Ranki,6 Richard J. Powell,5

diminished shedding of potentially antagonistic sol-Graham A. Hitman,1 John J. O’Shea,3

uble receptor. TNFR1-associated periodic syndromesand Daniel L. Kastner2,14

(TRAPS) establish an important class of mutations in1 Medical UnitTNF receptors. Detailed analysis of one such mutationSt. Bartholomew’s and the Royal London Hospitalsuggests impaired cytokine receptor clearance as aSchool of Medicine and Dentistrynovel mechanism of disease.Whitechapel, London E1 1BB, England

2 Genetics3 Lymphocyte Cell Biology4 Clinical Investigations Sections IntroductionArthritis and Rheumatism BranchNational Institute of Arthritis and Clinical and molecular genetic studies define several

Musculoskeletal and Skin Diseases heritable disorders that present as self-limited episodesBethesda, Maryland 20892-1820 of fever and serosal, synovial, or cutaneous inflammation5 Division of Molecular and Clinical Immunology (reviewed in Centola et al., 1998). Two of these illnesses,

familial Mediterranean fever (FMF, MIM249100) and theSchool of Clinical Laboratory Scienceshyperimmunoglobulinemia D with periodic fever syn-Queen’s Medical Centredrome (MIM260920), are inherited as autosomal reces-University of Nottinghamsive traits. The gene causing FMF (designated MEFV)Nottingham NG7 2UH, Englandwas recently identified by positional cloning (Interna-6 Department of Dermatologytional FMF Consortium, 1997; French FMF Consortium,and Divisions of Internal Medicine and Nephrology1997); it encodes a novel protein, denoted pyrin (alterna-Department of Medicinetively, marenostrin), that is expressed predominantly inHelsinki University Central Hospitalleukocytes and is homologous to a number of nuclearHelsinki 00029 HUCH, Finlandfactors. The gene mutated in the hyper-IgD syndrome7 Departments of Medicine and Rheumatologyis not linked to MEFV on chromosome 16 or to the heavyNational University of Irelandchain immunoglobulin locus (IgH) on chromosome 14Cork, Ireland(Drenth et al., 1994).8 Department of Clinical Genetics

The classification of the dominantly inherited periodicThe New Children’s Hospitalfever syndromes is much less well-defined. The mostWestmead, New South Wales 2124, Australiathoroughly characterized pedigree is a large Irish/Scot-9 Rheumatology and Clinical Immunology Unittish family with a periodic inflammatory condition thatUniversity of Vermont College of Medicinehas been termed familial Hibernian fever (FHF, MIMBurlington, Vermont 05405142680). A number of clinical features distinguish this10 Department of Epidemiologydisorder from FMF, including longer average durationUniversity of Texasof attacks, presence of conjunctivitis and periorbitalM. D. Anderson Cancer Centeredema, the distribution of cutaneous involvement, andHouston, Texas 77030less-pronounced response to colchicine prophylaxis11 Department of Medicine(Williamson et al., 1982; McDermott and Powell, 1997;Harvard Medical School

Boston, Massachusetts 0211512 Department of Cytogenetics and Molecular Genetics

13 These authors contributed equally to this work.Center for Medical Genetics 14 To whom correspondence should be addressed (e-mail: m.f.Women’s and Children’s Hospital mcdermott@mds.qmw.ac.uk [M. F. McD.], kastnerd@arb.niams.nih.North Adelaide, South Australia 5006 gov [D. L. K.]).

Cell134

McDermott et al., 1997). Mulley et al. (1998) have charac- et al., 1994; Peschon et al., 1998b), suggesting that theproinflammatory effects of TNF are mediated predomi-terized a large Australian pedigree of Scottish ancestry

with a similar condition, denoting it familial periodic fever nantly through the p55 receptor.Ligand-independent self-association of TNFR1 cyto-(FPF). Smaller pedigrees of other ethnic backgrounds

have also been reported. Although families exhibiting a plasmic domains is inhibited by a 60 kDa “silencer ofdeath domains” (SODD; Jiang et al., 1999). TNF actionsdominant mode of inheritance have several common

features, including febrile episodes that often last more may also be constrained in part by metalloprotease-dependent cleavage of membrane-bound TNF recep-than one week, there are clinical differences among

them, most notably the types of skin rash and the pres- tors (Porteu and Nathan, 1990; Crowe et al., 1995; Mull-berg et al., 1995), which can be triggered by a numberence or absence of amyloidosis. Thus, it has not been

clear whether the dominant periodic fevers represent of proinflammatory stimuli. Cleavage is thought to con-tribute to the clearance of TNFR1 and TNFR2 from theone or more mutations at a single locus, or result from

mutations in several genes. membrane, and produces a pool of soluble receptorsthat may attenuate the inflammatory response by com-Genome-wide searches for susceptibility loci have

been independently performed on the Irish/Scottish FHF peting with membrane-bound receptors (Engelmann etal., 1989). Mice homozygous for a targeted mutationfamily (with two other Irish families) as well as the Austra-

lian FPF family of Scottish descent. Disease loci in these in an enzyme that cleaves several membrane proteins(including TNF and TNFR2) have numerous develop-families were recently mapped to distal chromosome

12p (McDermott et al., 1998; Mulley et al., 1998), with mental defects (Peschon et al., 1998a), emphasizing thephysiologic importance of ectodomain shedding.FPF localized by recombination to the z19 cM between

D12S364 and D12S314, and FHF mapped to the z11 Given the central role of TNFR1 in inflammation, itslocation within the FHF/FPF interval, and the observa-cM between D12S77 and D12S93 (which is totally sub-

sumed within the FPF interval). These data suggest that tion of abnormal levels of soluble p55 in patients’ blood,we hypothesized that gain-of-function mutations in thisthe same locus may underlie disease susceptibility in

all of the families studied, and raise the possibility that gene might explain dominantly inherited periodic feverin at least some families. Here we report the identifica-this locus might also account for susceptibility in other

similar pedigrees. tion of six different missense coding mutations in sevenfamilies of varied ethnic backgrounds, including the fourBased on recent transcript maps (Deloukas et al.,

1998), there are a number of plausible positional candi- pedigrees recently linked to chromosome 12p13. Five ofthese mutations involved cysteines participating in di-date genes within the D12S77-D12S93 FHF/FPF interval,

including CD4, LAG-3, CD27, and the complement sulfide bonds in the first and second extracellular TNFR1domains, while the sixth substituted a methionine for agenes C1R and C1S. In addition, one of the two known

receptors for the proinflammatory cytokine tumor necro- highly conserved threonine adjacent to a cysteine in-volved in disulfide bonding. In considering mechanismssis factor (TNF) is encoded within the candidate region.by which these mutations might induce inflammation,This 55 kDa receptor (p55, TNFR1, CD120a; Loetscherwe evaluated several possibilities, including (1) in-et al., 1990; Schall et al., 1990) is widely expressed oncreased affinity of mutant TNFR1 for ligand; (2) constitu-a number of cell types. A second, 75 kDa receptor (p75,tive activation, possibly through the formation of inter-TNFR2, CD120b; Smith et al., 1990) is found predomi-molecular disulfide bonds between unpaired cysteinesnantly on leukocytes and endothelial cells and is en-in mutant receptors (Santoro et al., 1995); and (3) resis-coded on chromosome 1p.tance of mutant TNFR1 to the normal homeostatic ef-We took a particular interest in the TNFR1 gene be-fects of activation-induced cleavage. Analysis of leuko-cause of preliminary observations on the prototype FHFcytes from the three affected members of a family withfamily suggesting that between attacks FHF patientsa C52F mutation favors the third possibility.have reduced blood levels of a soluble form of TNFR1

relative to healthy controls, and that during attacks theseResultslevels increase much less than in other inflammatory

conditions. TNFR1 was also an attractive candidate be-cause of the range of biologic actions of TNF (reviewed Mutational Analyses

The seven families studied in the present report includein Vassalli, 1992). These pleiotropic activities includeincreased expression of adhesion molecules on leuko- the FPF family of Scottish ancestry recently linked to

chromosome 12p13 (Mulley et al., 1998), the prototypecytes and endothelial cell surfaces, induction of cytokinesecretion, leukocyte activation, host resistance to intra- Irish/Scottish Hibernian fever family and one large Irish

family from the FHF linkage study (McDermott et al.,cellular pathogens, angiogenesis, pyrexia, anemia, andcachexia. Both TNFR1 and TNFR2 belong to a family 1998), a Finnish family (Valimaki et al., 1985; Karenko et

al., 1992), and three small North American families ofof receptors with repeating cysteine-rich extracellularmotifs (Bazzoni and Beutler, 1996). TNF homotrimers Irish/English/German (Zaks and Kastner, 1997), Irish,

and French-Canadian ancestry. Of the four pedigreesinduce signaling by aggregation of membrane receptors(Engelmann et al., 1990). TNFR1 contains an z70 residue that had not previously undergone microsatellite analy-

sis, only the Finnish family was large enough to permit“death domain” that is actually a protein–protein interac-tion domain involved in signal transduction (Hsu et al., conclusive linkage studies. Genotypes for D12S356

yielded a maximal pairwise LOD score of 3.1 at u 5 01995). Knockout mice for TNFR1 are much more suscep-tible to challenge with intracellular bacterial pathogens in the Finnish family, and an 11-microsatellite haplotype

spanning the FHF/FPF candidate region segregatedand bacterial lipopolysaccharide than are wild-type con-trols (Pfeffer et al., 1993) or TNFR2 knockouts (Erickson with disease.

TNFR1 Mutations in Autoinflammatory Disease135

Figure 1. DNA Sequence Electropherograms Demonstrating Disease-Associated TNFR1 Mutations

A summary of the mutations, including the ethnicities of the families studied and the number of control chromosomes screened, is shown inthe upper panels. Electropherograms for individuals heterozygous for the mutations are shown in the middle panels, and homozygotes forthe normal alleles are shown in the lower panels. The relevant nucleotide is indicated with a down-arrow in the middle panels.

For each family, a single nucleotide substitution was In the second extracellular domain of TNFR1, twoidentified upon genomic sequencing of the coding re- mutations cause missense substitutions for the cysteinegion and splice junctions of the TNFR1 gene (Figure 1). at residue 88. In the Australian FPF family, a mutationFour mutations produced missense changes in the first at nt 349 in all seven of the available affected memberscysteine-rich extracellular domain of the TNFR1 protein. results in an arginine at this residue (C88R). In 4/4 af-A mutation leading to the substitution of arginine for fected members of the Finnish family, a G→A transitioncysteine at residue 30 (relative to the signal peptide at nt 350 produces a tyrosine at this residue (C88Y). Forcleavage site) was observed in the two affected mem- these latter two mutations, control chromosomes werebers of the Irish-American family (C30R). In 13 affected screened by restriction endonuclease digests, usingmembers analyzed from the prototype FHF family, a primers designed in such a way that the mutation eitherG→A transition results in the substitution of tyrosine for creates (C88R) or abolishes (C88Y) a restriction site incysteine at residue 33 (C33Y). In one branch of this the amplified DNA. Figure 2 illustrates the use of the TaqIfamily, three individuals reported to have periodic fevers assay for C88R in DNAs from affected and unaffecteddid not possess the substitution, but they also did not members of the Australian FPF family. When this assayshare the microsatellite haplotype present in all other was performed on 134 Northern European control chro-affected members, and the diagnosing clinician has not

mosomes, and when the HinP1I assay for C88Y waswitnessed the attacks of any of these three individuals.

performed on 122 Finnish control chromosomes, noneEight of eight affected members of the Irish FHF family,were positive for the mutation.and the one available member of the French-Canadian

Altogether, these six mutations were observed in 40family, had a mutation leading to the substitution ofof 43 symptomatic family members; the three variantmethionine for threonine at residue 50 (T50M). Two ofnegative individuals all belong to the same family andtwo additional members of the Irish family with mildmay represent genetic heterogeneity or misdiagnoses.symptoms also proved to have this mutation. Finally,The T50M and C88Y mutations were each observed inthe three affected members of the Irish/English/Germana single asymptomatic adult member of the Irish andfamily had a G→T transversion that causes the substitu-the Finnish family, respectively. The six mutations weretion of phenylalanine for cysteine at residue 52 (C52F).not observed in any of the other 40 healthy relativesNone of these changes were found in any of over 100studied, or in over 100 ethnically matched control chro-Northern European control chromosomes screened by

genomic sequencing. mosomes.

Cell136

Figure 2. TaqI Restriction Endonuclease Assayfor the C88R Mutation in an Australian Familyof Scottish Ancestry

The upper part of the figure depicts the pedi-gree of this family, adapted from Mulley etal. (1998). Filled symbols represent affectedfamily members; DNA was available fromnumbered individuals. Genomic DNA wasPCR-amplified with a primer set designedsuch that the C88R mutation creates a TaqIrestriction endonuclease cleavage site in theamplified DNA. Amplicons were digested withthe restriction enzyme, and electrophoresedon polyacrylamide gels. The lower part of thefigure depicts a photograph of the ethidiumbromide-stained gel from this family. Notethat all affected individuals have two bands,and that all unaffected individuals have oneband.

Soluble p55 Receptor Levels Are Diminished 127.4 pg/ml) and normal individuals (1002.1 6 261.8pg/ml). Nonpenetrant individuals and those with mildin the Blood of Individuals Bearing TNFR1

Extracellular Mutations symptoms all had low p55 levels, arguing that reducedsoluble p55 receptor levels are not simply a conse-The examination of TNFR1 as a positional candidate

had been encouraged by preliminary data suggesting quence of repeated episodes of inflammation, but di-rectly related to genotype. The p55 levels varied signifi-reduced levels of soluble p55 in the blood of affected

members of the prototype FHF family. We have ex- cantly among families (F17,43 5 3.09, p , 0.001), but thisfinding reflects variability only for the noncarriers (F15,18 5tended these serologic studies to a total of 27 mutation-

positive individuals with four different mutations (Table 3.47, p , 0.007), with no significant variability amongfamilies for mutation carriers (F3,23 5 2.3, p 5 0.11). Al-1). Statistical analyses showed a highly significant

(F1,43 5 94.3, p , 0.0001) difference in p55 levels between though the mean p55 level among mutation carriers dur-ing an attack (926.0 6 270.1 pg/ml) was significantlymutation carriers not experiencing attacks (541.6 6

Table 1. Soluble TNF Receptor Levels in Families with Dominantly Inherited Periodic Fever

Family (Mutation) p55 (TNFR1)a pb p75 (TNFR2)a pb

Prototype FHF (C33Y)Affected, between attacks (8) 591 6 115 ,0.025 2297 6 380 1.00Affected, during attacks (5) 889 6 284 2718 6 648Unaffected (7)c 899 6 142 2093 6 479

Irish (T50M)Affected, between attacks (11) 571 6 117 ,0.0001 —Unaffected (9)d 1273 6 248 —

Finnish (C88Y)Affected, between attacks (5) 402 6 113 ,0.0001 —Unaffected (11)d 785 6 103 —

Irish/English/German (C52F)e

Affected, between attacks (2)f 624 6 77 3009 6 1295Affected, during attacks (1)g 1111 6 259 3138 6 974Controls (9)h 1077 6 215 1994 6 293Active RA, SLE (2) 10,225 6 5324 5970 6 783

a Values are expressed as mean pg/ml 6 SD.b Statistical significance between affecteds not experiencing attacks and unaffecteds by the paired t test.c Two unaffected members of the same family, five healthy controls.d Unaffected members of the same family.e Statistics were not calculated for this family, because of the small sample size.f Two determinations for each of two individuals.g Four determinations for a single individual.h Unaffected, unrelated individuals.

TNFR1 Mutations in Autoinflammatory Disease137

Figure 3. Mutated Amino Acids Are at Highly Conserved Positions in the TNFR Superfamily

Several members of the TNF receptor family were aligned using a variation of BLAST that compares two sequences to each other. The firstten extracellular disulfide bonds predicted by crystallographic analysis of TNFR1 (Banner et al., 1993) are numbered and shown as linesconnecting conserved cysteine residues. The amino acids in which we have observed mutations in patients are marked with asterisks andare numbered.

greater (p , 0.02) than between attacks (541.6 6 127.4 Functional TNF Receptors Are Expressedon Leukocytes of Patientspg/ml), elevations were much less than in the control

patients with active rheumatoid arthritis or systemic lu- with the C52F MutationFigure 3 depicts the alignment of TNFR1 with severalpus erythematosus.

To evaluate the possibility that reduced soluble p55 other members of this receptor family, indicating thatfive of these mutations involve a conserved residuelevels might be the result of a more general deficit, solu-

ble p75 levels were also measured in two families. Levels shown crystallographically (Banner et al., 1993) to par-ticipate in disulfide bonding. Although they are notof p55 and p75 were not correlated (r 5 0.09, p 5 0.62),

and mean levels of p75 did not significantly vary be- points of receptor–ligand contact, the five cysteine sub-stitutions, and perhaps T50M as well, would significantlytween mutation carriers and noncarriers (p 5 0.64).

Moreover, levels of p75 were similar (p 5 0.8) between affect the structure of this molecule. If these changescaused ubiquitination and degradation, mutated TNFR1attacks (2439.2 6 623.6 pg/ml) and during attacks

(2788.2 6 604.5 pg/ml). might never reach the cell membrane, and mutations

Figure 4. TNF Receptors on Leukocytes fromPatients with the C52F Mutation

PMNs (A and B) and monocytes (C and D)were analyzed for CD120a (TNFR1) andCD120b (TNFR2) expression. Single cell sus-pensions of peripheral blood leukocytes (106

cells) were incubated with anti-CD120a-PE,anti-CD120b-PE, or isotype-matched IgG-PEas control (histogram 1), washed, and ana-lyzed by flow cytometry. Histograms 4, 5, and6 (in red) represent TNFR expression in thethree patients with the C52F mutation, andhistograms 2 and 3 (in green) show the ex-pression in two different control donors. (E)PMNs from patients with the C52F mutationbind TNF. Patient (histogram 3 and 5, in red)and control donor (histogram 2 and 4, ingreen) PMNs were incubated with a FITC-conjugated monoclonal antibody that recog-nizes both membrane-bound TNF and TNFbound to its receptors, or an isotype-matched FITC-IgG control (histogram 1).Cells were preincubated with (histogram 4and 5) or without (histogram 2 and 3) humanrecombinant TNF (1 mg/ml) for 1 hr at 48C,washed 3 times, and incubated with anti-TNFor control antibody for 30 min at 48C. After 3washes, cells were analyzed by flow cy-tometry.

Cell138

Figure 5. Leukocytes Bearing the C52F Mutation Have Defective Clearance of TNFR1 (CD120a)

TNFR1 Mutations in Autoinflammatory Disease139

on one allele might therefore result in the reduction in (Garcia et al., 1995), and exogenous soluble TNF recep-tor can ameliorate inflammatory disease in experimentalsoluble p55 noted above. Figure 4 presents flow cyto-

metric data relevant to this issue in the three affected animals (Mohler et al., 1993) and humans (Moreland etal., 1997). Impaired receptor clearance might thereforemembers of the family bearing the C52F mutation.

Somewhat unexpectedly, panels A and B indicate in- explain the autoinflammatory phenotype and the in-creased cell surface TNFR1 in C52F patients, as well ascreased surface TNFR1 (CD120a) but not TNFR2 (CD120b)

staining in polymorphonuclear leukocytes (PMNs) from the reduced soluble p55 blood levels.To study TNFR1 cleavage more directly, monocytesthe three individuals bearing this mutation (histograms

4–6) relative to normal controls (histograms 2 and 3). and PMNs from C52F patients and controls were treatedwith 4b-phorbol 12-myristate 13-acetate (PMA), a potentSimilar data for monocytes are presented in panels C

and D. inducer of the metalloproteases mediating receptorshedding (Mullberg et al., 1995). Part A of Figure 5 com-Figure 4E explores the capacity of C52F leukocytes

to bind TNF. Comparison of histograms 2 and 3 indicates pares CD120a (TNFR1, p55) staining of monocytes fromthree patients and two controls. In all three patients,that the total cell surface TNF (membrane plus receptor-

bound) was greater for the patient PMNs than normal unstimulated monocytes (histogram 2) had about 10-fold greater staining than the controls. Histogram 3 de-controls. When recombinant TNF was added and total

membrane TNF was detected with a nonneutralizing picts CD120a staining 10 min after PMA stimulation, andhistogram 4 depicts staining 1 hr after addition of PMA.antibody, similar increases in membrane TNF were ob-

served in patient (histogram 5 versus 3) and control In normal monocytes, there was marked diminution inmembrane CD120a at 10 min and nearly complete clear-(histogram 4 versus 2) PMNs, thus confirming the capac-

ity of receptors on the patients’ cells to bind ligand. ance by 1 hr. By contrast, there was virtually no reduc-tion in CD120a staining at either time point in the pa-Experiments not shown in the figure demonstrated that

C52F PMNs pretreated with blocking antibodies against tients’ monocytes.Part B of Figure 5 depicts the results of similar studiesTNFR2 bound biotinylated TNF at levels comparable to

controls. Although the autoinflammatory phenotype of on PMNs. Unstimulated patient cells had substantiallymore surface CD120a than controls, and addition ofthese patients would be consistent with a heightened

affinity of mutant receptors for their ligand, binding stud- PMA resulted in a rapid and complete reduction of stain-ing in the controls. Ten minutes after addition of PMAies of 125I-labeled TNF to PMNs demonstrated a Kd of

448 6 79 pM for three C52F patients, which was not there was minimal attenuation of staining in all threepatients; at the 1 hr time point, staining of the patients’significantly different from the value of 618 6 266 pM

measured in five controls. Although these figures proba- PMNs had diminished, but there were still substantiallevels of CD120a.bly represent the average affinity of several species of

mutant/wild-type receptor combinations, they do not Panel C summarizes the PMN data at the 10 min timepoint, sharply contrasting the minimal reductions insupport the hypothesis that inflammatory attacks are

caused by heightened affinity of receptor for TNF. CD120a in C52F patients with the substantial attenua-tion observed in healthy controls. Moreover, leukocytesWe also considered the possibility that the C52F mu-

tation could cause constitutive activation, perhaps by from patients with active systemic lupus erythematosusor rheumatoid arthritis, which expressed high levels ofpermitting intermolecular disulfide homodimerization

and ligand-independent activation (Santoro et al., 1995). TNFR1, cleared CD120a promptly. Panel D demon-strates that both C52F patients and controls had similarTo test this hypothesis, purified peripheral blood mono-

nuclear cells were stimulated with varying concentra- PMA-induced reductions in CD16 staining, thus rulingout a more general defect in clearance.tions of TNF, and supernatants were assayed for IL-6, a

TNF-responsive cytokine. Baseline IL-6 production was To determine whether the in vitro stimulation systemwould recapitulate the diminished soluble p55 levelsnot increased, and the dose–response for purified

mononuclear cells was similar to controls (data not observed in vivo, we performed ELISA assays for solublep55 on supernatants of PMA-stimulated leukocytes. Theshown), thus arguing against constitutive activation.results are shown in Figure 6. Shed p55 ectodomainlevels were markedly lower in supernatants of patientsImpaired Clearance of Membrane TNFR1

in Patients with the C52F Mutation relative to healthy controls at all time points (panel A)and over a broad range of PMA doses (panel B), despiteMembrane TNFR1 is regulated in part by metallopro-

tease-mediated cleavage. Shed receptors retain ligand- the fact that p55 was expressed at much higher levelson patients’ cells. Panel C shows several determinationsbinding activity and may have an anti-inflammatory

effect by competing for TNF binding with membrane re- of shed p55 receptor. Whereas cells from the C52F pa-tients had reduced levels of receptor shedding, patientsceptors (Engelmann et al., 1989). Moreover, transgenic

mice expressing high levels of a TNFR1 fusion protein with active autoimmune disease had increased PMA-induced TNFR1 shedding, thus reinforcing the point thatare resistant to several of the biological effects of TNF

Monocytes (A) and PMNs (B) were analyzed for CD120a expression, before (histogram 2) or after PMA activation (10 ng/ml) for 10 min(histogram 3) or 1 hr (histogram 4). Single cell suspensions (106 cells) were incubated with anti-CD120a-PE, or isotype-matched IgG-PE ascontrol (histogram 1), washed, and analyzed by flow cytometry. (C and D) Cells were stained with PE-coupled anti-CD120a (C) or FITC-coupledanti-CD16 (D) before and after PMA activation (10 min, 10 ng/ml) and analyzed by flow cytometry. Mean fluorescence intensity was measuredin control donors (green, open circles), C52F mutation patients (red, filled circles), or autoimmune patients (shaded diamonds).

Cell140

Figure 6. Leukocytes Bearing the C52F Mu-tation Have Impaired Shedding of TNFR1(p55) as Detected by ELISA

(A and D) Peripheral blood cells from controldonors (open circles) or C52F patients (filledcircles) were stimulated with PMA (100 ng/ml) for 5, 10, 30, and 60 min (A and D). (B andE) Cells from controls (open circles) or C52Fpatients (filled circles) were activated for 1 hrwith increasing doses of PMA. (C and F) Cellsfrom control donors (open circles), from C52Fpatients (filled circles), or from autoimmunepatients (shaded diamonds) were activatedfor 1 hr with PMA (10 ng/ml). Supernatantswere collected and soluble TNFR1 (p55; A–C)and TNFR2 (p75; D–F) were measured byELISA.

impaired clearance is not an inevitable consequence of their absence in healthy family members or in any of alarge panel of control chromosomes, all argue that thesesystemic inflammation.

Panels D–F present similar analyses for TNFR2, the substitutions are not benign polymorphisms. Moreover,we have demonstrated functional abnormalities in thep75 receptor. In contrast to p55, shedding of p75 was

greater in patient samples than in controls at 10 and 20 membrane-bound p55 TNF receptors of patients withthe C52F mutation, thereby suggesting a probable mecha-min after PMA stimulation (panel D). A dose–response

at 1 hr indicated nearly identical p75 receptor shedding nism by which this mutation causes disease.The autosomal dominant periodic fevers thereforebetween patient and control samples at all doses of

PMA (panel E). Panel F demonstrates that, unlike p55, represent a class of human disease shown to be causedby mutations in TNF receptors. Autoantibodies are notp75 shedding in C52F patients was consistently in the

high–normal range, at levels comparable to autoimmune a general feature of these illnesses or the recessivelyinherited FMF, and for this reason the term autoinflam-patients.matory is preferable to autoimmune in describing thesedisorders. Variation in ethnic background, pattern ofDiscussioncutaneous involvement, and presence of systemic amy-loidosis have created uncertainty about the etiology andIn this paper, we present compelling evidence that muta-

tions in the extracellular domains of TNFR1 can cause classification of the dominantly inherited periodic fevers.The identification of TNFR1 mutations in several familiesa dominantly inherited syndrome of episodic fever and

inflammation. In each of seven different families we have of disparate ethnic backgrounds and clinical pheno-types is a persuasive argument for a nomenclature thatdemonstrated a missense point mutation predicted to

have a profound effect on the structure of TNFR1 by unifies those families sharing this common pathogene-sis. We propose the descriptive acronym TRAPS (TNFdisrupting highly conserved intrachain disulfide bonds.

The identification of six independent disease-associ- receptor-associated periodic syndromes) to include allcases with documented TNFR1 mutations, while re-ated abnormalities in the same gene, the clustering of

these substitutions within two cysteine-rich motifs, and taining historical names (FHF, FPF, etc.) to denote indi-

TNFR1 Mutations in Autoinflammatory Disease141

vidual variants or for clinical diagnosis. The identification with the loss of a C-terminal cytoplasmic negative regu-latory domain. Atopy has been associated with codomi-of six independent receptor mutations is inconsistent

with a major founder effect in TRAPS. nant inheritance of an IL-4a receptor polymorphism,which may impair binding of intracellular negative regu-This report also suggests impaired cytokine receptor

clearance as a novel mechanism of human disease. In latory phosphatases (Hershey et al., 1997).Data from patients bearing the C52F TNFR1 mutationthe three affected members of a family with the C52F

mutation, we found markedly increased cell-surface suggest that the TRAP syndromes are likely to be dueto a mechanism different from those described above.TNFR1 relative to healthy controls. Stimulation with

PMA, which induces metalloprotease-mediated cleav- In the patients presented here, a mutation affecting thefirst extracellular domain impairs the normal regulatoryage of TNFR1, resulted in much less clearance of mem-

brane receptor in these patients than in controls, and processes of membrane clearance and ectodomainshedding, while the intracellular signaling domain ofdiminished shedding of receptor ectodomain into the

supernatant. Control experiments with CD16 and TNFR2 TNFR1 is intact. The C52F TNFR1 mutation appears tobe a novel type of dominant negative, in which “nega-refuted the possibility of a generalized defect in receptor

clearance, and studies of patients with active autoim- tive” refers not to a failure of signaling, but instead toa failure of shedding. It may serve as a paradigm notmune disease confirmed prior evidence for increased

receptor shedding with systemic inflammation (reviewed only for those TRAPS families in which we have notformally explored the mechanism of disease, but forin Aderka, 1996). Moreover, for the family with C52F

and others with different mutations, circulating levels of additional as yet undiscovered mutations in other mem-bers of the TNF receptor family. Mutations in Fas, an-TNFR1 between inflammatory episodes were lower than

those of healthy controls, and the increased levels ob- other member of the TNF receptor family, can cause adominantly inherited syndrome of lymphoproliferationserved during attacks were substantially less than would

be observed in other inflammatory diseases. and autoimmunity, but the mutations identified to datehave not involved this mechanism (Drappa et al., 1995;We therefore propose that the C52F mutation inter-

feres with normal homeostasis both by impairing post- Fisher et al., 1995; Rieux-Laucat et al., 1995).TRAPS patients can have disease-free intervals ofstimulatory TNF receptor clearance, and by reducing

levels of shed soluble receptor that can compete with months to years, suggesting that the mutated proteinfunctions adequately much of the time and requires anthe membrane form. TNF directly stimulates p55 shed-

ding from monocytes and induces receptor shedding important environmental triggering event for disease. Insome cases the factors associated with attacks fit wellfrom PMNs via other proinflammatory intermediates

(Porteu and Hieblot, 1994). Site-directed mutagenesis of with our understanding of TNF pathways, as when mildinfections or localized trauma provoke a prolonged sys-TNFR1 suggests that the most likely target of activation-

induced receptor cleavage is Asn-172–Val-173 (Gullberg temic inflammatory episode. For other frequently men-tioned inciting factors, such as psychological stress andet al., 1992), which is spatially remote from the mutations

described here. Possibly, TRAPS mutations cause con- physical exertion, or protective factors, such as preg-nancy, the connection with TNF is less well-established.formational changes in TNFR1 that interfere with metal-

loprotease binding. Alternatively, interchain disulfide Given that we have observed six different mutationsin seven families, it is premature to speculate on whetherbonds may form between unpaired cysteines, resulting

in steric inhibition of metalloprotease interactions, or phenotypic variation correlates with specific mutationsin TNFR1, or with other background genes segregatingnecessitating two proteolytic “hits” for clearance of

membrane TNFR1 homodimers. in these families. Particularly important is the questionof whether certain TNFR1 mutations predispose to sys-For certain other cytokine receptors, interchain disul-

fide bonds at specific residues can cause constitutive, temic amyloidosis, a potentially life-threatening compli-cation. The amyloidosis of TRAPS is caused by the tis-ligand-independent activation (Watowich et al., 1992;

Alexander et al., 1995; Santoro et al., 1995). Ligand- sue deposition of a product of the acute phase reactantserum amyloid A (SAA), which is produced by hepato-independent activation is unlikely in TRAPS, given the

episodic clinical course and our inability to observe cytes in response to TNF and IL-6. Arguing against adirect genotype–phenotype correlation is the fact thatspontaneous IL-6 production by C52F mononuclear

cells. Constitutive overexpression of human p75 TNFR2 although the C33Y and C52F mutations both disruptthe third disulfide bond in the first extracellular domain,in transgenic mice results in a ligand-independent lethal

wasting disease (Douni and Kollias, 1998). The present amyloidosis has developed in all adult members of theC52F family, but in only 1 of 21 affected members ofclinical and laboratory data are inconsistent with consti-

tutive TNFR1 activation in the C52F patients, though the family with the C33Y mutation. In contrast with muta-tions of the Fas gene (Fisher et al., 1995), TNFR1 muta-more subtle effects remain to be examined, and this

mechanism may be important in TRAPS patients with tions are nearly completely penetrant, with only 2/42 (5%)mutation-positive individuals reporting no symptoms.other mutations.

TRAPS also stands out as one of the few dominantly Observation of TRAPS patients will complement stud-ies of transgenic and knockout mice in deepening ourinherited disorders of cytokine receptors described to

date. Systemic mastocytosis can be caused by a gain- understanding of the inflammatory cascade. Particularlyintriguing are questions related to the predilection ofof-function cytoplasmic substitution in c-Kit, the recep-

tor for stem cell ligand, leading to constitutive autophos- TRAPS attacks for certain anatomic sites, such as theserosa, joints, and conjunctivae. A recent clinical analy-phorylation (Nagata et al., 1995). Dominantly inherited

erythrocytosis can result from truncating mutations in sis of the prototypic FHF family (C33Y) demonstratedthat 8/10 affected males had inguinal hernias, versusthe erythropoietin receptor (de la Chapelle et al., 1993),

Cell142

found, we treated family effects as a random component for subse-1/21 unaffected male family members (McDermott et al.,quent analysis of variance procedures. We tested whether there1997). One female member of the Irish C30R family de-were differences between mutation carriers and noncarriers usingveloped a severe neurologic illness in her childhood thata mixed effects analysis of variance design in which family member-

has persisted into adult life, a finding that may be related ship was treated as a random component and mutation status wasto neurologic abnormalities reported in TNF receptor treated as a fixed effect. To compare mutation carriers during and

between attacks, we used paired and two-sample t tests. For theknockout mice (Bruce et al., 1996). In contrast to TNFtwo-sample t test using p55 levels, interindividual variance duringtransgenic mice (Keffer et al., 1991), TRAPS patientsattacks was significantly greater than the variance between attacks,have not been observed to develop erosive arthritis.and so we used a t test assuming unequal variances. Where avail-Laboratory studies of leukocytes from TRAPS patientsable, replicate observations for an individual were averaged. Statisti-

will also extend our understanding of TNF-mediated sig- cal analyses were completed using the Statistical Analysis Systemnaling and apoptosis. Treatment of patients with newly (SAS).developed TNF inhibitors, such as the recombinant p75TNFR:Fc fusion protein (Moreland et al., 1997) or the Mutation Detection by Fluorescent SequencingcA2 chimeric monoclonal antibody (Elliott et al., 1994; Approximately 100 ng of genomic DNA template was used in PCRTargan et al., 1997), will further delineate the biologic reactions to amplify exons (Fuchs et al., 1992) and flanking intronic

sequences according to the supplier’s recommendations for Ampli-role of TNF in these patients, while providing a rationalTaq Gold (Perkin Elmer) or the Advantage-GC Genomic PCR Kittherapeutic approach to a condition for which there have(Clontech). The PCR primers were tailed with either 221 M13 for-been few options.ward, TGTAAAACGACGGCCAGT, or 228 M13 reverse, AGGAAA

Thus, the findings presented in this paper define TRAPS CAGCTATGACCAT. Forward (F) and reverse (R) oligonucleotidesas a new class of autoinflammatory disease, establish used to amplify exons were as follows (all sequences are given 59impaired cytokine receptor clearance as a novel patho- to 39): exon 1F, ACTTGGGACGTCCTGGACAGACC; exon 1R, GGA

AGGTGCCTCGCCCACCAG (193 bp); exon 2/3F, CCTCTCTTGATGGgenetic mechanism, and promise to extend our under-TGTCTCC; exon 2/3R, CTGACTCTCCTGCCTGTGC (575 bp); exonstanding of the TNF cascade in health and disease. In4/5F, GGTATGTCAGGAAGGGGATG; exon 4/5R, CAGTGCCCAGClight of the central role TNF plays in inflammation, it isAGCTCTC (546 bp); exon 6/7F, CAATGGTAGGGCCTCTGTTC; exon

likely that additional TNF receptor mutations, or muta- 6/7R, CACCCATGTCCTCCCTCAC (430 bp); exon 8/9F, TGATGCTTTtions of molecules upstream or downstream in the TNF CTTTCTTTTTC; exon 8/9R, CAGCCTCCTCGTCTCCAG (577 bp).pathway, may be found in a range of other inflammatory Exon 10 was amplified in two overlapping fragments: exon 10AF,

GGACCACGAGAGGCGGAG; exon 10AR CCATCCCAGCAGCTCdiseases.CAGCGT (291 bp); exon 10BF, ATCGATCGGCTGGAGCTGCAG; exon10BR, ACCCCTCCTTTCCAGAAAAA (296 bp). PCR products were

Experimental Procedures sequenced by dye primer chemistry (Amersham) on an ABI 377automated sequencer as described (International FMF Consortium,

Patients and DNA Samples 1997).Clinical details and pedigrees of five of the seven families studiedin this report have been previously described (Karenko et al., 1992;

Restriction Endonuclease Assays for TNFR1 MutationsMcDermott et al., 1997; Zaks and Kastner, 1997; family B in McDer-Restriction endonuclease assays were developed for five of themott et al., 1998; Mulley et al., 1998). Samples were also obtainedmutations reported here. For each assay, 50–100 ng of genomicfrom one affected member of a French-Canadian family and twoDNA was amplified in a total volume of 10 ml in the presence of 1.5affected siblings from an Irish-American family. A total of 85 individu-mmol MgCl2 PCR buffer, 0.04 U AmpliTaq Gold, and 1 pmol of eachals were studied, 41 of whom had been judged to be affected byof the appropriate PCR primers. For C33Y, the exon 2/3F primer wasclinical criteria (McDermott et al., 1997; Mulley et al., 1998), and anused with TCCACTTGCCCCTACCTTTGTAG; for T50M, the primersadditional 2 of whom had mild symptoms. Consent was obtainedwere TGTGTTCTCACCCGCAGCCTAAC (F) and TCTCACACTCCCTfrom study participants after approval of the human experimentationGCAGTAC (R); for C52F, the T50M forward primer was used withcommittees at each institution. DNA was extracted from whole bloodthe exon 2/3R primer; for C88R, the primers were TTACAGAGACAor transformed lymphocytes by standard techniques.CACACTTAGG (F) and CACGGTGTCCCGGTCCACTGTTC (R); andfor C88Y, the C88R forward primer was used with CACGGTGTCCCGGenotyping and Linkage AnalysisGTCCACTGCG. PCR began with 958C for 10 min, followed by 30DNAs from the Finnish, Irish, and prototype FHF family were geno-cycles of 958C for 30 s, 558C (T50M) or 608C (C33Y, C52F, C88R,typed for 11 microsatellites spanning the FHF/FPF candidate region.and C88Y) for 30 s, and 728C for 30 s, with the final extension atFrom telomere to centromere, these were D12S341, D12S93,728C for 10 min. The amplified products were digested for at leastD12S99, D12S356, D12S374, TNFR55, CD4, D12S397, D12S1695,2 hr with 10 units of the appropriate restriction endonuclease. Re-D12S77, and D12S358. PCR amplifications were performed as de-striction digests were as follows: C33Y, ScaI (mutation creates site);scribed (McDermott et al., 1998); sequences of oligonucleotides andT50M, SnaBI (mutation abolishes site); C52F, PstI (mutation abol-specific conditions used to amplify these markers are available fromishes site); C88R, TaqI (mutation creates site); and C88Y, HinP1Ithe Genome Database, except for TNFR55, for which the primers(mutation abolishes site). Samples were loaded on 4%–20% poly-were 59 CACAGTCTCTCCTGTATCAT 39 (forward) and 59 CCTTCacrylamide gels (Novex) and stained with ethidium bromide.TGAATTTCTCACC 39 (reverse). LOD scores were calculated as de-

scribed (McDermott et al., 1998).

Cell Activation and Flow Cytometric AnalysisBlood cells or neutrophils sedimented through Ficoll-Hypaque gra-Measurement of Soluble TNF Receptors

Soluble TNFR1 (p55) and TNFR2 (p75) were measured in plasma or dients were washed and incubated at 378C for 10 min or 1 hr inHanks’ balanced salt solution (HBSS) containing 1% fetal calf serumculture supernatants by solid-phase ELISA (R&D Systems, Minneap-

olis). For studies of plasma receptor levels, simple descriptive stud- (FCS), with or without PMA (Sigma). Cells were washed withHBSS/1% FCS at 48C and erythrocytes were lysed using ammoniumies were completed for p55 and p75 levels to document approximate

normality. The correlation between p55 and p75 levels was assessed chloride potassium (ACK) lysing buffer (BioWhittaker). Expressionof membrane TNFR1 and TNFR2 were detected by flow cytometryby the Pearson product–moment correlation coefficient. To test

whether p55 and p75 levels varied among mutation carriers between using phycoerythrin (PE)-conjugated anti-CD120a and anti-CD120bmonoclonal antibodies (Caltag) and isotype controls (Becton Dickin-attacks, we used a fixed-effects analysis of variance procedure.

Because no significant evidence for family-specific effects was son); membrane CD16 was detected with fluorescein isothiocyanate

TNFR1 Mutations in Autoinflammatory Disease143

(FITC)-conjugated monoclonal antibodies (Becton Dickinson). Mono- Centola, M., Aksentijevich, I., and Kastner, D.L. (1998). The heredi-tary periodic fever syndromes: molecular analysis of a new familycytes were identified by staining with conjugated anti-CD14 mono-of inflammatory diseases. Hum. Mol. Genet. 7, 1581–1588.clonal antibodies (Becton Dickinson), and lymphocyte, monocyte,

and neutrophil populations were analyzed after flow cytometric gat- Crowe, P.D., Walter, B.N., Mohler, K.M., Otten-Evans, C., Black,ing. Cells were incubated with conjugated antibodies or isotype R.A., and Ware, C.F. (1995). A metalloprotease inhibitor blocks shed-controls for 30 min at 48C and washed 3 times with phosphate- ding of the 80-kD TNF receptor and TNF processing in T lympho-buffered saline containing 0.5% bovine serum albumin (BSA). Sam- cytes. J. Exp. Med. 181, 1205–1210.ples were analyzed on a FACS Calibur flow cytometer (Becton Dick- de la Chapelle, A., Traskelin, A.L., and Juvonen, E. (1993). Truncatedinson). erythropoietin receptor causes dominantly inherited benign human

erythrocytosis. Proc. Natl. Acad. Sci. USA 90, 4495–4499.

Deloukas, P., Schuler, G.D., Gyapay, G., Beasley, E.M., Soderlund,TNF Binding AssaysC., Rodriguez-Tome, P., Hui, L., Matise, T.C., McKusick, K.B., Beck-Serial dilutions of 125I-TNF (2000–62.5 pM; NEN Life Science) weremann, J.S., et al. (1998). A physical map of 30,000 human genes.added to 2.2 3 106 cells in a final volume of 200 ml (RPMI 1640,Science 282, 744–746.1% BSA) and incubated for 4 hr on ice. Background binding was

measured in the presence of a 200-fold molar excess of cold ligand. Douni, E., and Kollias, G. (1998). A critical role of the p75 tumorBound ligand was determined by centrifuging the cells through a necrosis factor receptor (p75TNF-R) in organ inflammation indepen-dibutylphthalate/bis(2-ethylhexyl)phthalate (60:40) cushion and cut- dent of TNF, lymphotoxin a, or the p55TNF-R. J. Exp. Med. 188,ting off the tip of the tube to count the bound radioactivity. 1343–1352.

Drappa, J., Vaishnaw, A.K., Sullivan, K.E., Chu, J.-L., and Elkon,K.B. (1995). Fas gene mutations in the Canale-Smith syndrome, anTNF In Vitro Stimulation Assaysinherited lymphoproliferative disorder associated with autoimmu-Peripheral blood mononuclear cells were isolated by centrifugationnity. N. Engl. J. Med. 335, 1643–1649.on Ficoll-Hypaque gradients. The cells were washed and resus-Drenth, J.P.H., Mariman, E.C.M., van der Velde-Visser, S.D., Ropers,pended at a concentration of 106/ml in RPMI 1640 supplementedH.H., van der Meer, J.W.M., and the International Hyper-IgD Studywith 10% heat-inactivated FCS, penicillin/streptomycin, and 2 mMGroup (1994). Location of the gene causing hyperimmunoglobuli-L-glutamine; 1 ml cultures were established in 24-well platesnemia D and periodic fever syndrome differs from that for familial(Costar). Cells were cultured with medium alone or with TNF (9, 18,Mediterranean fever. Hum. Genet. 94, 616–620.90, 180, 900, and 1800 pg/ml) for 48 hr. Cell-free supernatants were

collected by centrifugation and analyzed by ELISA for IL-6 (R&D Elliott, M.J., Maini, R.N., Feldmann, M., Long-Fox, A., Charles, P.,Systems). Bijl, H., and Woody, J.N. (1994). Repeated therapy with monoclonal

antibody to tumour necrosis factor alpha (cA2) in patients with rheu-matoid arthritis. Lancet 344, 1125–1127.AcknowledgmentsEngelmann, H., Aderka, D., Rubinstein, M., Rotman, D., and Wallach,D. (1989). A tumor necrosis factor-binding protein purified to homo-We are grateful to the families for agreeing to participate in thegeneity from human urine protects cells from tumor necrosis factorstudy. M. McD. was in receipt of a Wellcome Trust Travel Fellowshiptoxicity. J. Biol. Chem. 264, 11974–11980.for part of the work; M. A. was supported by a Postdoctoral ResearchEngelmann, H., Holtmann, H., Brakebusch, C., Avni, Y.S., Sarov, I.,Exchange Grant of the Max Kade Foundation. Technical assistanceNophar, Y., Hadas, E., Leitner, O., and Wallach, D. (1990). Antibodieswas provided by K. Jarvinen and M. Raatikainen. We are gratefulto a soluble form of a tumor necrosis factor (TNF) receptor haveto S. Nedospasov (Frederick, MD and Moscow) and R. TuretskayaTNF-like activity. J. Biol. Chem. 265, 14497–14504.(Moscow) for unpublished TNFR55 microsatellite oligonucleotide

sequences, to J. Tuomilehto (Helsinki) for control Finnish DNA sam- Erickson, S.L., de Sauvage, F.J., Kikly, K., Carver-Moore, K., Pitts-ples, to Tracy Thiel Costello (Houston) for assistance with linkage Meek, S., Gillett, N., Sheehan, K.C.F., Schreiber, R.D., Goeddel, D.V.,analysis, and to G. Gallagher and I. Ragoussis for helpful discussion. and Moore, M.W. (1994). Decreased sensitivity to tumour-necrosisSupported in part by the Research Advisory Committee of the Spe- factor but normal T-cell development in TNF receptor-2-deficientcial Trustees, the Royal London Hospitals NHS Trust; the Jones mice. Nature 372, 560–563.Charitable Trust, Nottingham, UK; the National Health and Medical Fisher, G.H., Rosenberg, F.J., Straus, S.E., Dale, J.K., Middelton,Research Council of Australia; the Medical Research Funds of L.A., Lin, A.Y., Strober, W., Lenardo, M.J., and Puck, J.M. (1995).Tampere and Helsinki University Hospitals, Tampere and Helsinki, Dominant interfering Fas gene mutations impair apoptosis in a hu-Finland; and the U.S. National Institutes of Health (grants AR44422 man autoimmune lymphoproliferative syndrome. Cell 81, 935–946.and GM52607). French FMF Consortium (1997). A candidate gene for familial Medi-

terranean fever. Nat. Genet. 17, 25–31.

Fuchs, P., Strehl, S., Dworzak, M., Himmler, A., and Ambros, P.F.Received February 17, 1999; revised March 11, 1999.(1992). Structure of the human TNF receptor 1 (p60) gene (TNFR1)and localization to chromosome 12p13. Genomics 13, 219–224.

ReferencesGarcia, I., Miyazaki, Y., Araki, K., Araki, M., Lucas, R., Grau, G.E.,Milon, G., Belkaid, Y., Montixi, C., Lesslauer, W., et al. (1995).Aderka, D. (1996). The potential biological and clinical significanceTransgenic mice expressing high levels of soluble TNF-R1 fusionof the soluble tumor necrosis factor receptors. Cytokine Growthprotein are protected from lethal septic shock and cerebral malaria,

Factor Rev. 7, 2331–2340.and are highly sensitive to Listeria monocytogenes and Leishmania

Alexander, W.S., Metcalf, D., and Dunn, A.R. (1995). Point mutations major infections. Eur. J. Immunol. 25, 2401–2407.within a dimer interface homology domain of c-Mpl induce constitu-

Gullberg, U., Lantz, M., Lindvall, L., Olsson, I., and Himmler, A. (1992).tive receptor activity and tumorigenicity. EMBO J. 14, 5569–5578. Involvement of an Asn/Val cleavage site in the production of a solu-Banner, D.W., D’Arcy, A., Janes, W., Gentz, R., Schoenfeld, H.-J., ble form of a human tumor necrosis factor (TNF) receptor. Site-Broger, C., Loetscher, H., and Lesslauer, W. (1993). Crystal structure directed mutagenesis of a putative cleavage site in the p55 TNFof the soluble human 55 kd TNF receptor–human TNFb complex: receptor chain. Eur. J. Cell Biol. 58, 307–312.implications for TNF receptor activation. Cell 73, 431–445. Hershey, G.K.K., Friedrich, M.F., Esswein, L.A., Thomas, M.L., andBazzoni, F., and Beutler, B. (1996). The tumor necrosis factor ligand Chatila, T.A. (1997). The association of atopy with a gain-of-functionand receptor families. N. Engl. J. Med. 334, 1717–1725. mutation in the a subunit of the interleukin-4 receptor. N. Engl. J.

Med. 337, 1720–1725.Bruce, A.J., Boling, W., Kindy, M.S., Peschon, J., Kraemer, P.J.,Carpenter, M.K., Holtsberg, F.W., and Mattson, M.P. (1996). Altered Hsu, H., Xiong, J., and Goeddel, D.V. (1995). The TNF receptorneuronal and microglial responses to excitotoxic and ischemic brain 1-associated protein TRADD signals cell death and NF-kB activa-

tion. Cell 81, 495–504.injury in mice lacking TNF receptors. Nat. Med. 2, 788–794.

Cell144

International FMF Consortium (1997). Ancient missense mutations Rieux-Laucat, F., Le Deist, F., Hivroz, C., Roberts, I.A.G., Debatin,in a new member of the RoRet gene family are likely to cause familial K.M., Fischer, A., and de Villartay, J.P. (1995). Mutations in FasMediterranean fever. Cell 90, 797–807. associated with human lymphoproliferative syndrome and autoim-

munity. Science 268, 1347–1349.Jiang, Y., Woronicz, J.D., Liu, W., and Goeddel, D.V. (1999). Preven-tion of constitutive TNF receptor 1 signaling by silencer of death Santoro, M., Carlomagno, F., Romano, A., Bottaro, D.P., Dathan,domains. Science 283, 543–546. N.A., Grieco, M., Fusco, A., Vecchio, G., Matoskova, B., Kraus, M.H.,

et al. (1995). Activation of RET as a dominant transforming gene byKarenko, L., Pettersson, T., and Roberts, P. (1992). Autosomal domi-germline mutations of MEN2A and MEN2B. Science 267, 381–383.nant ‘Mediterranean fever’ in a Finnish family. J. Int. Med. 232,

365–369. Schall, T.J., Lewis, M., Koller, K.J., Lee, A., Rice, G.C., Wong, G.H.W.,Gatanaga, T., Granger, G.A., Lentz, R., Raab, H., et al. (1990). Molec-Keffer, J., Probert, L., Cazlaris, H., Georgopoulos, S., Kaslaris, E.,ular cloning and expression of a receptor for human tumor necrosisKioussis, D., and Kollias, G. (1991). Transgenic mice expressingfactor. Cell 61, 361–370.human tumour necrosis factor: a predictive genetic model of arthri-

tis. EMBO J. 10, 4025–4031. Smith, C.A., Davis, T., Anderson, D., Solam, L., Beckmann, M.P.,Jerzy, R., Dower, S.K., Cosman, D., and Goodwin, R.G. (1990). ALoetscher, H., Pan, Y.-C.E., Lahm, H.-W., Gentz, R., Brockhaus,receptor for tumor necrosis factor defines an unusual family of cellu-M., Tabuchi, H., and Lesslauer, W. (1990). Molecular cloning andlar and viral proteins. Science 248, 1019–1023.expression of the human 55 kd tumor necrosis factor receptor. Cell

61, 351–359. Targan, S.R., Hanauer, S.B., van Deventer S.J.H., Mayer, L., Present,D.H., Braakman, T., DeWoody, K.L., Schaible, T.F., Rutgeerts, P.J.,McDermott, E.M., and Powell, R.J. (1997). Colchicine therapy infor the Crohn’s Disease cA2 Study Group (1997). A short-term studyfamilial Hibernian fever. In Familial Mediterranean Fever, E. Sohar,of chimeric monoclonal antibody cA2 to tumor necrosis factor a forJ. Gafni, and M. Pras, eds. (London and Tel Aviv: Freund), pp. 24–27.Crohn’s disease. N. Engl. J. Med. 337, 1029–1035.McDermott, E.M., Smillie, D.M., and Powell, R.J. (1997). The clinical

spectrum of familial Hibernian fever: a 14-year follow-up study of Valimaki, M., Anttila, P.M., Pentikainen, P.J., Tornroth, T., Maury,the index and extended family. Mayo Clin. Proc. 72, 806–817. P., and Nyman, J.A. (1985). Valimerenkuumetta Itameren rannalla?

(Mediterranean fever in Finland?) Duodecim 101, 2428–2432.McDermott, M.F., Ogunkolade, B.W., McDermott, E.M., Jones, L.C.,Wan, Y., Quane, K.A., McCarthy, J., Phelan, M., Molloy, M.G., Powell, Vassalli, P. (1992). The pathophysiology of tumor necrosis factors.R.J., et al. (1998). Linkage of familial Hibernian fever to chromosome Annu. Rev. Immunol. 10, 411–452.12p13. Am. J. Hum. Genet. 62, 1446–1451. Watowich, S.S., Yoshimura, A., Longmore, G.D., Hilton, D.J., Yoshi-Mohler, K.M., Torrance, D.S., Smith, C.A., Goodwin, R.G., Stremler, mura, Y., and Lodish, H.F. (1992). Homodimerization and constitutiveK.E., Fung, V.P., Madani, H., and Widmer, M.B. (1993). Soluble tumor activation of the erythropoietin receptor. Proc. Natl. Acad. Sci. USAnecrosis factor (TNF) receptors are effective therapeutic agents in 89, 2140–2144.lethal endotoxemia and function simultaneously as both TNF carri- Williamson, L.M., Hull, D., Mehta, R., Reeves, W.G., Robinson, B.H.,ers and TNF antagonists. J. Immunol. 151, 1548–1561. and Toghill, P.J. (1982). Familial hibernian fever. Quart. J. Med. 51,Moreland, L.W., Baumgartner, S.W., Schiff, M.H., Tindall, E.A., 469–480.Fleischmann, R.M., Weaver, A.L., Ettlinger, R.E., Cohen, S., Koop- Zaks, N., and Kastner, D.L. (1997). Clinical syndromes resemblingman, W.J., Mohler, K., et al. (1997). Treatment of rheumatoid arthritis familial Mediterranean fever. In Familial Mediterranean Fever, E.with a recombinant human tumor necrosis factor receptor (p75)-Fc Sohar, J. Gafni, and M. Pras, eds. (London and Tel Aviv: Freund),fusion protein. N. Engl. J. Med. 337, 141–147. pp. 211–215.Mullberg, J., Durie, F.H., Otten-Evans, C., Alderson, M.R., Rose-John, S., Cosman, D., Black, R.A., and Mohler, K.M. (1995). A metal-loprotease inhibitor blocks shedding of the IL-6 receptor and thep60 TNF receptor. J. Immunol. 155, 5198–5205.

Mulley, J., Saar, K., Hewitt, G., Ruschendorf, F., Phillips, H., Colley,A., Sillence, D., Reis, A., and Wilson, M. (1998). Gene localizationfor an autosomal dominant familial periodic fever to 12p13. Am. J.Hum. Genet. 62, 884–889.

Nagata, H., Worobec, A.S., Oh, C.K., Chowdhury, B.A., Tannenbaum,S., Suzuki, Y., and Metcalfe, D.D. (1995). Identification of a pointmutation in the catalytic domain of the protooncogene c-kit in pe-ripheral blood mononuclear cells of patients who have mastocytosiswith an associated hematologic disorder. Proc. Natl. Acad. Sci. USA92, 10560–10564.

Peschon, J.J., Slack, J.L., Reddy, P., Stocking, K.L., Sunnarborg,S.W., Lee, D.C., Russell, W.E., Castner, B.J., Johnson, R.S., Fitzner,J.N., et al. (1998a). An essential role for ectodomain shedding inmammalian development. Science 282, 1281–1284.

Peschon, J.J., Torrance, D.S., Stocking, K.L., Glaccum, M.B., Otten,C., Willis, C.R., Charrier, K., Morrissey, P.J., Ware, C.B., and Mohler,K.M. (1998b). TNF receptor-deficient mice reveal divergent roles forp55 and p75 in several models of inflammation. J. Immunol. 160,943–952.

Pfeffer, K., Matsuyama, T., Kundig, T.M., Wakeham, A., Kishihara,K., Shahinian, A., Wiegmann, K., Ohashi, P.S., Kronke, M., and Mak,T.W. (1993). Mice deficient for the 55 kd tumor necrosis factor recep-tor are resistant to endotoxic shock, yet succumb to L. monocyto-genes infection. Cell 73, 457–467.

Porteu, F., and Hieblot, C. (1994). Tumor necrosis factor induces aselective shedding of its p75 receptor from human neutrophils. J.Biol. Chem. 269, 2834–2840.

Porteu, F., and Nathan, C. (1990). Shedding of tumor necrosis factorreceptors by activated human neutrophils. J. Exp. Med. 172,599–607.