Mutant tumor necrosis factor receptor associated with tumor necrosis factor receptor–associated periodic syndrome is altered antigenically and is retained within patients' leukocytes

Authors


Abstract

Objective

To investigate the effect of mutations in tumor necrosis factor receptor superfamily member 1A (TNFRSF1A) in TNFR-associated periodic syndrome (TRAPS) on the binding of anti-TNFRSF1A monoclonal antibodies (mAb), and to investigate the subcellular distribution of mutant versus wild-type (WT) TNFRSF1A in patients with TRAPS.

Methods

HEK 293 cells transfected with WT and/or mutant TNFRSF1A were used to investigate the interaction of anti-TNFRSF1A mAb with the WT and mutant proteins. Monoclonal antibodies that differentially bound to C33Y TNFRSF1A were used to investigate the distribution of WT and mutant TNFRSF1A in TRAPS patients with the C33Y mutation.

Results

We identified a mAb whose binding to TNFRSF1A was completely abolished by the C33Y or C52F TRAPS-associated mutations, whereas other mutations (T50M, C88Y, R92Q) had lesser effects on the binding of this mAb. A different mAb was found to bind efficiently to all of the mutant forms of TNFRSF1A examined as well as to the WT receptor. Exploitation of the differential binding properties of these mAb indicated that mutant (as distinct from WT) TNFRSF1A showed abnormal intracellular retention in the neutrophils of TRAPS patients with the C33Y mutation, with little if any expression of mutant TNFRSF1A on the cell surface or as soluble receptor in plasma.

Conclusion

TRAPS-associated mutant TNFRSF1A has an antigenically altered structure and shows abnormal retention in the leukocytes of patients with TRAPS, which is consistent with previous findings from in vitro and transgenic model systems. This is consistent with a misfolded protein response contributing to the pathophysiology of TRAPS.

Tumor necrosis factor receptor (TNFR)–associated periodic syndrome (TRAPS; OMIM 142680) is a hereditary autoinflammatory periodic fever syndrome. It is associated with autosomal-dominant mutations in the ectodomain of the 55-kd TNFR (TNFR superfamily 1A [TNFRSF1A], TNFR1, TNFR p55/p60, CD120a) (1–4). More than 50 different mutations have been identified in patients with TRAPS (refs. 2 and5, and online at: http://fmf.igh.cnrs.fr/infevers). These mutations are likely to have significant structural effects on the conformation and stability of the receptor's ectodomain (6).

Defective neutralization of TNF by soluble TNFRSF1A (sTNFRSF1A) may contribute to the pathophysiology of TRAPS, due to reduced shedding of the receptor (1, 7–10), although TNF levels are not elevated in TRAPS patients with the C33Y mutation (11). We proposed that mutant TNFRSF1A may perform ligand-independent signaling as a consequence of misfolding and intracellular aggregation, which may also contribute to the pathophysiology of TRAPS (6, 12). Consistent with this, we observed in transfection systems that most mutant forms showed reduced cell surface expression and TNF binding and increased intracellular retention, while retaining signaling functions (6, 9, 12). Other investigators have reported similar abnormalities of TNFRSF1A mutants in transfected cells (13, 14). In particular, Lobito et al (14) showed that mutant TNFRSF1A did not interact with wild-type (WT) TNFRSF1A and was largely retained in the endoplasmic reticulum (ER) and formed disulfide-bonded oligomers. Those investigators also demonstrated intracellular retention of mutant TNFRSF1A in vivo in a “knock-in” mouse expressing the T50M mutant form of TNFRSF1A (14).

The degree of abnormal behavior of mutant TNFRSF1A correlates with the predicted structural effects of the mutations, with substitutions of cysteine residues involved in the disulfide bonds showing the greatest abnormalities (6, 9, 12, 14); these cysteine mutations can also have particularly severe clinical consequences (2, 3). Functional consequences of TNFRSF1A mutations that have been reported include either increased (15) or decreased (13, 14, 16, 17) activation of NF-κB and reduced induction of apoptosis (13, 14, 16, 17).

A single amino acid substitution can affect the stability and conformation of the ectodomain region of TNFRSF1A (6, 14, 15), and it would be expected that the most profound structural effects would be in the immediate vicinity of the mutation. In the studies presented here, we defined an anti-TNFRSF1A monoclonal antibody (mAb) whose binding to TNFRSF1A was completely abolished by the C33Y or C52F mutation in the first cysteine-rich domain (CRD1). (In the WT protein, a disulfide bond is formed between C33 and C52.) Other mutations had lesser effects on the binding of this mAb. In contrast, a different mAb was shown to bind efficiently to all of the mutant forms of TNFRSF1A examined as well as to the WT receptor. We were then able to use the differential binding properties of these 2 mAb to define the relative distribution of WT and mutant TNFRSF1A in the cells and plasma of TRAPS patients with the C33Y mutation. (Because the mutations are dominant, patients with TRAPS are heterozygous for mutant and WT TNFRSF1A.)

PATIENTS AND METHODS

Samples from patients with TRAPS and healthy control subjects.

Peripheral blood samples from related TRAPS patients with a C33Y mutation in TNFRSF1A and from healthy control subjects were collected in Vacutainers containing EDTA (BD Biosciences, Oxford, UK). All participants gave informed consent, and the study was approved by the Nottingham (UK) Local Research Ethics Committee and the University of Nottingham Medical School Research Ethics Committee. Plasma samples were stored until used for detection of sTNFRSF1A by enzyme-linked immunosorbent assay (ELISA). On the day of collection, leukocytes were stained for surface and intracellular expression of TNFRSF1A (see below).

Production and stable transfection of recombinant DNA clones of WT and mutant TNFRSF1A.

The production of recombinant DNA clones of WT and mutant TNFRSF1A in the plasmid vector pcDNA4/TO, encoding the intact full-length receptor or lacking the cytoplasmic signaling region (Δsig) due to a stop codon at residue 215, has been described previously (9, 12). Clones were produced encoding the WT receptor and receptors with the mutations C33Y, T50M, or C52F in CRD1 and C88Y or R92Q in CRD2. Transfection of Tetracycline-Regulated Expression HEK 293 cells with these DNA clones, which was performed to produce stably transfected cell lines that express recombinant TNFRSF1A (WT, mutant, or a combination of both) under doxycycline control, has also been described previously (9, 12).

Production and transient transfection of recombinant DNA clones of WT and mutant TNFRSF1A-GFP(topaz).

Mutant and WT TNFRSF1A coding sequences (lacking the terminator codon) were inserted into the plasmid pDEST-TOPAZ, using the Gateway recombination system (Invitrogen, Paisley, UK). Plasmid pDEST-TOPAZ is a green fluorescent protein (GFP) topaz variant of pEGFP-N1, with the C1 Gateway conversion cassette inserted at the Sma I site within the multiple cloning site. These constructs produced TNFR fusions, with topaz GFP as a C-terminal fusion.

HEK 293 cells were grown in Dulbecco's modified Eagle's medium containing 100 IU/ml penicillin, 100 μg/ml streptomycin, 10 mM HEPES buffer, 2 mML-glutamine, and 10% fetal calf serum (FCS) in 6-well plates, to 90–95% confluence. TNFRSF1A-GFP(topaz) plasmid DNA (8 μg) was diluted in Opti-MEM I (Invitrogen, Paisley, UK) and combined with 20 μl of Lipofectamine 2000 (LF2K) (Invitrogen), also diluted in Opti-MEM I. A 250-μl volume of DNA–LF2K was added to each well of the culture plate, as appropriate; the plate was then gently mixed and incubated at 37°C in a 5% CO2 atmosphere for 48 hours.

Production of lysates of transfected cells.

HEK 293 cells transiently transfected with TNFRSF1A-GFP(topaz) or stably transfected with TNFRSF1A, and untransfected control cells were harvested, washed, and resuspended at 106 cells per 50 μl of cell lysis buffer containing 50 mM Tris (pH 7.8), 150 mM NaCl, 1% Igepal CA-630 nonionic detergent (Sigma-Aldrich, Poole, UK), and 1 complete protease inhibitor tablet (Roche Diagnostics, Lewes, UK) per 10 ml of lysis buffer. Cells were incubated for 10 minutes on ice and then centrifuged at 1,000g for 10 minutes to pellet the nuclei. Supernatants were diluted at a ratio of 1:20 in phosphate buffered saline (PBS), and an ELISA was performed to detect TNFRSF1A.

ELISAs for the detection of TNFRSF1A.

The presence of TNFRSF1A in lysates of HEK 293 cells or in human plasma samples was detected as described previously (9), using the DuoSet TNFRSF1A detection kit (R&D Systems, Abingdon, UK). The wells of ELISA plates were coated with 4 μg/ml of the mouse monoclonal anti-TNFRSF1A DuoSet capture antibody (DCA). In modifications of the protocol, either mouse monoclonal anti-TNFRSF1A MAB225 (4 μg/ml) (R&D Systems) or rabbit polyclonal anti-GFP (1:1,000 dilution) (Abcam, Cambridge, UK) was used as the capture antibody in place of DCA; in all other respects, the protocols were identical. When using DCA or MAB225 as the capture antibody, a standard curve of the TNFRSF1A standard supplied with the DuoSet kit was applied to give concentrations in picograms per milliliter. When using anti-GFP as the capture antibody, a standard was not available and, therefore, optical density readings at 450 nm are reported.

Detection of surface and intracellular TNFRSF1A in HEK 293 cells.

Staining of transfected HEK 293 cells to detect TNFRSF1A, using phycoerythrin (PE)–conjugated MAB225 (25 μg/ml), was performed as previously described (6, 9, 12). For detection of TNFRSF1A by unlabeled DCA (5 μl of a 25-μg/ml concentration) or unlabeled MAB225 (5 μl of a 25-μg/ml concentration), fluorescein isothiocyanate (FITC)–labeled secondary goat anti-mouse IgG was used (Dako, Ely, UK); all incubations with the primary and secondary antibodies were for 30 minutes. Surface-expressed TNFRSF1A was detected on viable cells, and intracellular (plus surface) TNFRSF1A was detected in saponin-permeabilized cells. To investigate competition of the unlabeled MAB225 and DCA antibodies with PE-conjugated MAB225, the cells were coincubated with either 5 μl of a 25-μg/ml concentration of unlabeled DCA or MAB225 and 5 μl of a 25-μg/ml concentration of PE-conjugated mAb 225. Staining was detected using an EPICS XL flow cytometer (Beckman Coulter, High Wycombe, UK).

Detection of surface and intracellular TNFRSF1A in peripheral blood neutrophils.

For the detection of surface TNFRSF1A, 500 μl of OptiLyse C (Beckman Coulter) was added to 100 μl of peripheral whole blood collected in EDTA. The sample was mixed and then incubated for 10 minutes at room temperature. The cells were centrifuged at 300g for 5 minutes, resuspended in PBS containing 0.5% bovine serum albumin and 0.1% sodium azide (PBA), and centrifuged again. Cells were incubated for 45 minutes on ice in the dark with either 25 μg/ml of unlabeled DCA or MAB225, or were left untreated. Cells were washed twice in PBA and subsequently incubated with 5 μl of the FITC-labeled secondary antibody goat anti-mouse IgG. Cells were washed twice with PBA, resuspended in 0.5% formaldehyde fixative, and analyzed by flow cytometry.

For the detection of intracellular TNFRSF1A, 100 μl of peripheral blood was treated with OptiLyse C, as described above. The cells were then washed in PBA and subsequently fixed in 4% formaldehyde fixative for 5 minutes at room temperature. The cells were first washed with PBA, then with saponin buffer (PBA containing 0.04% saponin; Sigma-Aldrich), and finally with saponin buffer containing 10% FCS. Cells were incubated for 45 minutes on ice in the dark with either 25 μg/ml unlabeled DCA or MAB225, or with nothing at all. Cells were washed twice in saponin buffer and subsequently incubated, as before, with 5 μl of the FITC-labeled secondary antibody goat anti-mouse IgG. The cells were washed twice with saponin buffer, resuspended in 0.5% formaldehyde fixative, and analyzed by flow cytometry.

RESULTS

Differential mAb binding to C33Y TNFRSF1A in ELISA.

In ELISAs using either anti-GFP or MAB225 anti-TNFRSF1A as the capture antibody, TNFRSF1A was detected at similar levels in the lysates of cells transiently transfected with either WT or C33Y TNFRSF1A-GFP(topaz) (Table 1). In contrast, the anti-TNFRSF1A DCA bound WT TNFRSF1A but not C33Y TNFRSF1A (Table 1). This indicated that the epitope recognized by the DCA mAb in WT TNFRSF1A is absent from C33Y TNFRSF1A, but that the epitope for MAB225 is conserved between WT and C33Y TNFRSF1A.

Table 1. Results of ELISA for detection of differential binding of C33Y TNFRSF1A-GFP(topaz) by different capture monoclonal antibodies*
Cell lineAntibody used for TNFRSF1A capture
Protocol 1Protocol 2
Anti-GFPDCAMAB225DCA
  • *

    HEK 293 cells were transiently transfected with recombinant wild-type or C33Y TNFRSF1A-GFP(topaz) or were left untransfected. After 48 hours of culture, lysates of the cells were produced and assayed for TNFRSF1A-GFP(topaz) by capture enzyme-linked immunosorbent assay (ELISA). The capture antibody used to coat the wells of the ELISA plate was anti–green fluorescent protein (anti-GFP), DuoSet capture antibody (DCA), or monoclonal antibody (mAb) MAB225. In each case, the detection antibody was the DuoSet ELISA kit anti–tumor necrosis factor receptor superfamily 1A (anti-TNFRSF1A) detection mAb. ND = not determined.

  • Optical density.

  • Picograms of TNFRSF1A per milliliter.

Untransfected0.0030NDND
Wild-type0.659810.5364.5245.2
C33Y0.7530434.314.1

Differential binding of MAB225 and DCA to mutant forms of TNFRSF1A expressed in transfected cells.

The interaction of MAB225 and DCA with HEK 293 cells stably transfected with WT full-length TNFRSF1A or various full-length TNFRSF1A mutants was investigated. Immunofluorescence analysis was performed on permeabilized cells, because most mutant forms of recombinant full-length TNFRSF1A are retained in the cytoplasm, with little, if any, cell surface expression (6, 12–14). Using MAB225 as the detection antibody, high levels of all mutant forms of TNFRSF1A, as well as WT, were observed (Figure 1A). DCA similarly detected high levels of WT and R92Q TNFRSF1A but showed essentially no interaction with either the C33Y or C52F TNFRSF1A proteins (Figure 1A). DCA also showed slightly less interaction with T50M and C88Y TNFRSF1A than did MAB225 (Figure 1A).

Figure 1.

Differential binding of monoclonal antibody MAB225 and DuoSet capture antibody (DCA) to mutant forms of tumor necrosis factor receptor superfamily 1A (TNFRSF1A) expressed in transfected cells. A and B, HEK 293 cells stably transfected with wild-type (WT) or mutant full-length TNFRSF1A (A) or dually transfected with WT or mutant and WT TNFRSF1A lacking the cytoplasmic signaling region (Δsig) (B) were induced with doxycycline to express recombinant TNFRSF1A. Cells were permeabilized, stained with either MAB225 or DCA and fluorescein isothiocyanate–labeled secondary goat anti-mouse IgG (solid lines in A), and analyzed by flow cytometry. Broken lines represent secondary antibody alone. C, HEK 293 cells stably transfected with WT full-length TNFRSF1A were induced with doxycycline to express recombinant TNFRSF1A, permeabilized, and stained with phycoerythrin (PE)–conjugated MAB225, alone or in the presence of unlabeled MAB225 or DCA, and analyzed by flow cytometry.

We previously produced HEK 293 cell lines dually transfected with WT and mutant TNFRSF1A, to provide stable expression of heterozygous (WT/mutant) TNFRSF1A, as observed in cells from patients with TRAPS (9). Recombinant TNFRSF1A constructs lacking the cytoplasmic signaling region (Δsig) were used for these transfections (9). Staining of these cell lines, following permeabilization to allow detection of intracellular TNFRSF1A, showed very similar binding of MAB225 and DCA to the WT/WT and R92Q/WT dual transfectants, but lower binding of DCA than MAB225 to the C33Y/WT and C52F/WT transfectants; these observations are consistent with DCA binding to the WT but not the C33Y or C52F Δsig TNFRSF1A construct (Figure 1B). Lower binding of DCA than MAB225 was also observed with the T50M/WT and C88Y/WT dual transfectants (Figure 1B). (Cells within the C33Y/WT cell line that are completely negative for DCA binding, giving a peak coincident with that of the secondary antibody alone, are revertant cells that have lost recombinant WT Δsig TNFRSF1A expression.)

In view of the high levels of binding of MAB225 to all of the TNFRSF1A mutants examined here, including C33Y and C52F, the epitope for MAB225 is unlikely to be in close proximity to that for DCA. In order to confirm this, we investigated whether MAB225 and DCA compete for binding to WT TNFRSF1A. Monoclonal antibody MAB225 directly conjugated with PE was incubated with permeabilized HEK 293 cells expressing WT full-length TNFRSF1A in the presence of unlabeled MAB225 or DCA. As expected, reduced binding of PE-conjugated MAB225 was observed in the presence of unlabeled MAB225, whereas unlabeled DCA did not reduce binding of PE-conjugated MAB225 (Figure 1C), indicating that the epitopes for the 2 mAb do not overlap.

Interaction of MAB225 and DCA with combinations of WT and C33Y TNFRSF1A.

Because MAB225 interacts effectively with both WT and C33Y TNFRSF1A, whereas DCA interacts with WT but not with C33Y TNFRSF1A, the relative interactions of these mAb with a sample containing both WT and C33Y TNFRSF1A should correlate with the relative amounts of the WT and C33Y proteins. In particular, the MAB225:DCA binding ratio should increase as the ratio of C33Y to WT TNFRSF1A increases. Lysates of HEK 293 cells stably expressing either WT or C33Y recombinant full-length TNFRSF1A were assayed for TNFRSF1A by capture ELISA, both separately and combined in the ratios indicated in Figure 2 (maintaining a constant amount of lysate), with either MAB225 or DCA as the capture antibody. As shown in Figure 2A, similar amounts of TNFRSF1A were detected by MAB225 at various ratios of the WT and C33Y TNFRSF1A preparations (including WT or C33Y alone). In contrast, DCA detected decreasing amounts of TNFRSF1A as the proportion of C33Y relative to WT TNFRSF1A increased (Figure 2A). Thus, expressing these TNFRSF1A concentrations as the MAB225:DCA binding ratio gave an increasing value with increasing proportions of C33Y (Figure 2B).

Figure 2.

Interaction with MAB225 versus DCA correlates with the relative expression of WT and C33Y TNFRSF1A. A, TNFRSF1A levels. Lysates of stably transfected HEK 293 cells expressing WT or C33Y recombinant full-length TNFRSF1A were assayed by enzyme-linked immunosorbent assay (ELISA) for TNFRSF1A, either separately or combined in various ratios (maintaining a constant amount of lysate), using MAB225 or DCA as the capture antibody. In each case, the detection antibody was the DuoSet ELISA kit anti-TNFRSF1A detection monoclonal antibody. B, The MAB225:DCA TNFRSF1A detection ratio. See Figure 1 for other definitions.

Detection of WT, but not mutant, sTNFRSF1A in the plasma of TRAPS patients with the C33Y mutation.

Soluble TNFRSF1A is present in the blood as a consequence of cleavage from the surface of activated cells (18) and/or exosomal secretion (19). Plasma samples from TRAPS patients with the C33Y mutation and from normal healthy control subjects were assayed by capture ELISA for sTNFRSF1A, using MAB225 or DCA as the capture antibody. Because patients are heterozygous, if both WT and C33Y TNFRSF1A are present as sTNFRSF1A in their plasma, the MAB225:DCA binding ratio would be higher than that in plasma samples from normal control subjects, which contain only WT sTNFRSF1A. However, Figure 3 shows that there was no significant difference in the MAB225:DCA binding ratios between the patient and control samples (P = 0.1321 by unpaired t-test). This indicates that WT sTNFRSF1A is present in blood from patients with TRAPS, but that C33Y sTNFRSF1A is either not present in the blood or is present at much lower concentrations compared with the WT protein.

Figure 3.

TNFRSF1A detection ratio. The MAB225:DCA binding ratios determined by enzyme-linked immunosorbent assay in plasma samples from patients with TNFR-associated periodic syndrome with a C33Y mutation and those from control subjects are shown. Bars show the mean. No significant differences were observed by unpaired t-test. See Figure 1 for definitions.

Cellular expression of C33Y versus WT TNFRSF1A.

The ratio of MAB225:DCA binding to cellular TNFRSF1A should also correlate with the relative expression of C33Y and WT TNFRSF1A. Consistent with this, we observed (using flow cytometry) that the MAB225:DCA binding ratio on C33Y/WT Δsig TNFRSF1A dually transfected cells was almost twice as high for detection of cytoplasmic plus surface TNFRSF1A compared with staining of surface receptor alone, whereas this was not the case for WT/WT Δsig TNFRSF1A–transfected cells. This indicates that, relative to WT Δsig TNFRSF1A, a higher proportion of the C33Y protein is retained in the cytoplasm of the HEK 293 dual transfectants than is expressed on the cell surface, which is consistent with our previous findings in single transfectants (6, 12).

Staining of neutrophils by MAB225 and DCA was then investigated using blood samples from TRAPS patients with the C33Y mutation and normal healthy control subjects. This was performed on nonpermeabilized cells and permeabilized cells, to detect cell surface and cytoplasmic (as well as surface) TNFRSF1A expression, respectively. For staining of permeabilized neutrophils, the MAB225:DCA binding ratios were significantly higher for patients than for control subjects (P = 0.0226 by unpaired t-test) (Figure 4). For surface staining of neutrophils, however, there was no significant difference in these ratios between patients and control subjects (P = 0.4833 by unpaired t-test) (Figure 4). This indicates that C33Y TNFRSF1A (as well as WT TNFRSF1A) is present in the cytoplasm of cells from patients, but that C33Y TNFRSF1A shows little, if any, cell surface expression.

Figure 4.

Cytoplasmic, but not surface, expression of C33Y mutant tumor necrosis factor receptor superfamily 1A (TNFRSF1A) in neutrophils from patients with TNFR-associated periodic syndrome (TRAPS). Peripheral blood leukocytes from TRAPS patients with the C33Y mutation and normal healthy control subjects were left unpermeabilized (non-perm.) for surface staining or were permeabilized for cytoplasmic plus surface staining. Cells were stained with either MAB225 or DuoSet capture antibody (DCA) and fluorescein isothiocyanate–labeled secondary goat anti-mouse IgG and analyzed by flow cytometry. The mean fluorescence intensity (MFI) of staining of neutrophils by MAB225 or DCA was determined, and the MFI of staining by secondary antibody alone was subtracted. (Neutrophils were defined by forward and side scatter of light in the flow cytometer.) MAB225:DCA binding ratios are shown. Bars show the mean. Significant differences were determined by unpaired t-test.

DISCUSSION

Absence of the epitope recognized by the anti-TNFRSF1A DCA in C33Y and C52F TNFRSF1A mutants is strong evidence for the structural disruption of TNFRSF1A caused by TRAPS-associated mutations. The conformation of the DCA epitope seems to be particularly dependent on the disulfide bond between C33 and C52, because the neighboring T50M mutation has much less of an effect on DCA binding (Figures 1A and B). Localization of the DCA epitope to CRD1 is supported by the observation that the abrogation of a disulfide bond in CRD2 by the C88Y mutation has only a small effect on DCA binding (Figures 1A and B). Information was not available on the precise epitope of the DCA mAb, but the manufacturer's data sheet (R&D Systems) notes that TNF does not inhibit its interaction with TNFRSF1A, which binds to CRD2 and CRD3 of the receptor (20); this is consistent with DCA binding to CRD1. The fact that the C88Y mutation has any effect on DCA binding is consistent with our molecular modeling studies indicating that TRAPS-associated mutations affect the full length of the TNFRSF1A ectodomain (6). Lobito et al also highlighted the gross structural effects of TRAPS-associated mutations, since even noncysteine mutations appear to disrupt intramolecular disulfide bonds, resulting in disulfide-bonded oligomerization of mutant receptors (14).

The R92Q mutation, in contrast, had no effect on DCA binding (Figures 1A and B). This suggests that this mutation causes minimal disruption to the TNFRSF1A structure, which was also indicated by the results of our molecular modeling studies (6). It is also consistent with our own observations and those of other investigators that, in transfected cell lines, R92Q behaves very similarly to WT TNFRSF1A in terms of cell surface expression and TNF binding (6, 9, 12, 14). We have also observed relatively normal expression of R92E and R92K TNFRSF1A mutants in transfected cells (6), whereas R92P shows highly abnormal behavior (6, 14), presumably due to the strong structural influence of the proline residue. These findings of molecular studies correlate with the low penetrance and milder clinical phenotype associated with R92Q TNFRSF1A, which also occurs in 1–2.5% of the general population (7, 17, 21). It has also been associated with rheumatoid arthritis (7) and vascular diseases, i.e., atherosclerosis (22) and deep vein thrombosis in Beçhet's disease (23).

The high levels of MAB225 binding to all of the CRD1 and CRD2 TNFRSF1A mutants that we examined, as well as its lack of competitive inhibition by DCA (Figures 1A–C), suggest that its epitope lies membrane proximal to CRD2. Information was not available on the epitope of MAB225, but it competes with TNF for binding to the receptor (according to the manufacturer's data sheet), suggesting that its epitope may be within CRD3. This could be investigated by examining the effects of mutations in CRD3 on MAB225 binding; it would also be interesting to screen peptide phage display libraries to define the contact amino acid residues of DCA and MAB225 (24).

Studies of the MAB225:DCA binding ratio that we undertook with plasma and leukocytes from TRAPS patients with the C33Y mutation and unrelated healthy control subjects indicated that, in contrast to WT, C33Y TNFRSF1A is retained in the cytoplasm of neutrophils with little, if any, cell surface expression or release into the plasma as sTNFRSF1A (Figures 3 and 4). In other words, predominantly or only WT TNFRSF1A is expressed as the cell surface receptor and is present as soluble receptor in patients with TRAPS, as in healthy individuals. This is consistent with previous studies by our group and other investigators, in which it was observed that cell lines transfected with full-length TNFRSF1A expressed WT full-length TNFRSF1A on the surface to some extent as well as in the cytoplasm, whereas most mutant forms of full-length TNFRSF1A were expressed only in the cytoplasm, with very little or no cell surface expression (6, 12–14). This corroborates the use of these in vitro transfectants as model systems for investigating the molecular pathophysiology of TRAPS. It is also consistent with findings in the T50M TNFRSF1A “knock-in” mouse as an in vivo model of TRAPS (14). Furthermore, as observed in cell lines dually transfected with mutant and WT TNFRSF1A (9, 14) (Figure 4), the C33Y TNFRSF1A expressed in patients' neutrophils does not appear to grossly affect the cell surface expression or extracellular release of WT TNFRSF1A.

Our findings raise questions about the mechanism of the TNFRSF1A shedding defect observed in the leukocytes of some patients with TRAPS (1, 7, 8) if only WT TNFRSF1A is available on the cell surface to undergo activation-induced cleavage. The intracellular retention of mutant receptor could simply mean that less receptor is available on the surface to be cleaved, although leukocytes from TRAPS patients with the C52F mutation were reported to have normal or even increased surface expression of TNFRSF1A (1). The mutant TNFRSF1A may also have other effects on cells that down-regulate receptor release, e.g., effects on metalloproteinase activity (25) or aminopeptidase regulator of TNFRSF1A shedding 1 (26). Alternatively, because TNFRSF1A can be secreted as full-length receptor (19) as well as cleaved from the cell surface, it is possible that mutant receptor cannot be secreted because of its abnormal intracellular retention. However, regardless of the mechanism leading to the low levels of sTNFRSF1A commonly observed in the circulation of patients with TRAPS, these low levels are likely to cause suboptimal neutralization of TNF, as originally proposed (1).

TNFRSF1A mutants may have a number of effects that contribute to the pathophysiology of TRAPS (27), including TNF-dependent effects, such as the “shedding hypothesis” discussed above (1, 10), and the TNF-independent effects of TNFRSF1A mutants proposed by our group and other investigators (6, 12, 14, 15). The latter proposal is consistent with the intracellular retention of mutant TNFRSF1A in patients' neutrophils described here, which raises the possibilities of abnormal localization and aggregation of mutant receptors that might lead to inappropriate triggering of signaling pathways. (It is worth emphasizing that C33Y TNFRSF1A is, indeed, expressed in the cells of patients, which is consistent with the missense mutation allowing the transcription of functional messenger RNA, whereas a nonsense mutation might lead to RNA decay.)

The intracellular site(s) of accumulation of mutant TNFRSF1A in patients' cells remain to be determined. Cells transfected with recombinant TNFRSF1A can develop cytosolic aggregates of the receptor, although these may be, at least partly, an artefact of receptor overexpression (12–14). Evidence of retention in the ER has also been obtained (12, 14). Thus, inappropriate TNF-independent signaling by intracellular mutant TNFRSF1A could involve interactions of the receptors' death domains (or other cytoplasmic domains) within cytosolic aggregates formed if ER-associated protein degradation is incomplete (6, 12). Alternatively, it could involve ER stress responses as a consequence of the ER-associated misfolded proteins, although Lobito et al did not find evidence for activation of either the unfolded protein response or for the ER overload response (14). A clearer definition of the intracellular location of mutant TNFRSF1A in patients' cells would be facilitated by generating a mutant-specific mAb.

Overall, the results presented in this study demonstrate that TNFRSF1A with the C33Y mutation has an abnormal structure and shows abnormal retention in the leukocytes of patients with TRAPS, which is consistent with findings in model systems. This implicates a misfolded protein response in the pathophysiology of TRAPS, and indicates that it is a member of the growing family of protein conformational disorders (28). Indeed, the intracellular retention and spontaneous signaling proposed to occur with TRAPS-associated mutants of TNFRSF1A is analogous to other diseases in which mutated proteins are implicated (28, 29).

AUTHOR CONTRIBUTIONS

Dr. Todd had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Todd, Powell, Tighe.

Acquisition of data. Radford, Daffa, Bainbridge.

Analysis and interpretation of data. Todd, Radford, Daffa, Powell, Tighe.

Manuscript preparation. Todd, Radford, Powell, Tighe.

Statistical analysis. Todd.

Acknowledgements

We are grateful to Drs. Elizabeth McDermott and Elizabeth Drewe for obtaining the samples from the patients.

Ancillary