To investigate the effect of mutations in tumor necrosis factor receptor superfamily 1A (TNFRSF1A) on the ability of the receptors to be cleaved from the cell surface upon stimulation. The mutations we studied are associated with clinically distinct forms of TNF receptor–associated periodic syndrome (TRAPS). We also investigated different cell types within the same form of TRAPS.
The shedding of TNFRSF1A in response to stimulation with phorbol myristate acetate was assessed in leukocytes and dermal fibroblasts from patients with C33Y TRAPS, and in HEK 293 cell lines stably transfected with constructs containing wild-type TNFRSF1A and/or TNFRSF1A mutants identified in TRAPS patients.
The shedding of TNFRSF1A differed between cell types within the same form of TRAPS. In particular, dermal fibroblasts, but not leukocytes, from C33Y TRAPS patients demonstrated reduced shedding of TNFRSF1A. Shedding of both wild-type and mutant TNFRSF1A from the transfected HEK 293 cells showed minor differences, but was in all cases induced to a substantial extent.
Differences in TNFRSF1A shedding are not purely a function of the TNFRSF1A structure, but are also influenced by other features of genetic makeup and/or cellular differentiation. It is unlikely that a defect in TNFRSF1A shedding per se can fully explain the clinical features that are common to TRAPS patients with different TNFRSF1A mutations.
Tumor necrosis factor receptor–associated periodic syndrome (TRAPS; MIM 142680), which is characterized by recurrent fevers and certain inflammatory symptoms, is associated with autosomal-dominant mutations in the gene encoding the 55-kd tumor necrosis factor receptor (TNFRSF1A; also known as TNFR1, CD120a, and p55 TNFR). More than 30 different mutations have been reported (1–15).
The syndrome presented in the prototypical TRAPS family was originally termed familial Hibernian fever (FHF) (16, 17). Individuals with FHF experience febrile attacks, abdominal pain, localized myalgia, large, painful, erythematous skin lesions, conjunctivitis, periorbital edema, and inguinal hernias (17, 18). The associated mutation results in the substitution of cysteine by tyrosine at position 33 (C33Y) of the TNFRSF1A first cysteine-rich domain (CRD1) (1).
Both TNFRSF1A and the 75-kd TNF receptor (TNFRSF1B; also known as TNFR2, CD120b, and p75 TNFR) exist in soluble forms that are produced by cleavage by metalloproteinases upon cell activation (19). In common with patients bearing other TNFRSF1A mutations, the affected members of the C33Y prototype TRAPS family exhibit low levels of serum TNFRSF1A that rise to only just within the normal range during febrile attacks, unlike patients with, for example, rheumatoid arthritis, in whom soluble TNFRSF1A (sTNFRSF1A) can reach 10 times normal levels (1, 2, 20). Consistent with these in vivo observations, peripheral blood leukocytes from TRAPS family members with a C52F mutation were found to express higher than normal levels of TNFRSF1A and to exhibit little, if any, shedding of their surface TNFRSF1A when stimulated with phorbol myristate acetate (PMA), whereas TNFRSF1B did not share these abnormalities (1). This TNFRSF1A shedding defect led to the hypothesis that there might not be enough sTNFRSF1A to neutralize the free TNF, thus resulting in the continuous inflammatory stimulus that is notable in TRAPS (1). In addition, low levels of sTNFRSF1A are seen in all families with TRAPS, regardless of the mutation present (1). However, not all the TRAPS-related TNFRSF1A mutations result in defective receptor shedding by leukocytes (4, 14), suggesting that this may not be the sole cause of the low levels of sTNFRSF1A, and that other pathophysiologic mechanisms may also be involved in generating the disease phenotype.
In order to further elucidate the possible association of TNFRSF1A shedding with the pathophysiology of TRAPS, we investigated the shedding of surface TNFRSF1A from leukocytes and dermal fibroblasts from TRAPS patients with the C33Y mutation and from normal controls. Dermal fibroblasts were considered relevant to study because the patients demonstrate skin lesions, and fibroblasts are a potentially important source of inflammatory cytokines. Furthermore, in order to compare the shedding of TNFRSF1A with different mutations under conditions where all other potential genetic and cellular variables are constant, we produced HEK 293 cloned cell lines that stably expressed either wild-type (WT) and/or a single mutant recombinant TNFRSF1A in the absence of TNFRSF1B. Mutants with the following single–amino acid substitutions were produced as truncated forms, representing the extracellular and transmembrane regions but lacking the cytoplasmic signaling domain (Δ-sig): C33Y, T50M, C52F, C88Y, and R92Q. C33Y, T50M, and C52F are mutations within the CRD1 of TNFRSF1A; C88Y and R92Q are mutations in CRD2. We used these truncated forms of WT and mutant TNFRSF1A to assess receptor shedding in response to stimulation, because these do not affect cell viability when expressed, whereas expression of the full-length transfected receptor can induce apoptosis (ref. 21 and Todd I, et al: unpublished observations). Each of the single–amino acid substitutions used to generate the mutant cell lines has been previously described (1, 4).
MATERIALS AND METHODS
Treatment of leukocytes with PMA to measure sTNFRSF1A production.
Peripheral blood samples from 7 related C33Y TRAPS patients and 7 unrelated normal controls were collected into tubes containing EDTA. Fully informed consent was given by the patients; the controls consisted of anonymized routine blood samples.
The plasma was removed by centrifugation at 800g for 8 minutes. The remaining cells were washed by centrifugation at 350g for 5 minutes in RPMI 1640 medium (Sigma, Poole, UK) supplemented with 1% fetal calf serum (FCS; Harlan SeraLab, Loughborough, UK), 10 mM HEPES, 100 units/ml of penicillin, 100 μg/ml of streptomycin, and 2 mML-glutamine (Sigma). Cells were resuspended in RPMI 1640 and 1% FCS to the original volume of blood collected. One-milliliter aliquots of the plasma-free blood were incubated with or without 100 ng/ml of PMA (Sigma) for 10 minutes at 37°C. The samples were then centrifuged at 350g for 5 minutes, and the supernatant was removed and stored at –20°C prior to enzyme-linked immunosorbent assay (ELISA). The supernatants were analyzed with sTNFRSF1A and sTNFRSF1B Quantikine ELISAs (R&D Systems Europe, Abingdon, UK) by following the manufacturer's protocol.
Isolation of fibroblasts from skin.
Skin biopsy samples (<6 mm) were obtained from 2 male cousins with C33Y TRAPS (ages 46 and 54 years) and 2 unrelated sex- and age-matched control subjects (ages 44 and 51 years, respectively). Dermal fibroblasts were isolated from the biopsy samples according to the method of Barker and Clothier (22). Biopsies were taken from the center of a lesion in patient 1 and from the lesion-free skin of patient 2. Patient 1 was experiencing fever and myalgia at the time of biopsy. Patient 2 was experiencing intermittent attacks but did not have fever or myalgia at the time of biopsy. Both patients had elevated C-reactive protein levels. Both patients had experienced attacks since infancy. Fully informed consent was given by the patients and the controls.
Fibroblasts were maintained in Dulbecco's modified Eagle's medium (DMEM; Invitrogen, Paisley, UK) supplemented with 10 mM HEPES, 100 units/ml of penicillin, 100 μg/ml of streptomycin, 2 mML-glutamine, and 5–10% FCS at a concentration of 1–4 × 104 cells/ml of culture medium. The cells were passaged when 70–80% confluent, using 0.05% trypsin–0.02% EDTA (Sigma) to detach the adherent cells. To allow the cell lines to become reestablished after the same number of passages, samples of the cells were stored periodically in liquid N2. Although the cell lines studied were maintained at the same passage status, they are referred to in their original pairing (i.e., fibroblasts from patient 1/control 1 were established on the same day; those from patient 2/control 2 were established on the same day). Genetic analysis of the patients' fibroblasts confirmed the C33Y mutation.
TNFR shedding in response to PMA.
The fibroblasts were passaged at 0.5 × 104 cells in 0.5 ml of medium per well of 24-well culture plates (Costar, Cambridge, MA) for 5 days. The culture medium was replaced with fresh medium, and the cells were incubated for up to 48 hours with or without PMA (10 ng/ml). The culture supernatant was harvested, centrifuged at 3,000g for 5 minutes to remove any cell debris, then stored at –20°C prior to ELISA. A nontoxic dye, alamarBlue (Serotec, Oxford, UK) was used to assess cell viability without the need to disturb the attached cells (23). The reduction of the dye was detected with a Cytofluor fluorometer (Millipore, Watford, UK) at an excitation wavelength of 530 nm (±12.5 nm) and an emission wavelength of 590 nm (±10 nm). The supernatants were analyzed for sTNFRSF1A and sTNFRSF1B by ELISAs.
ELISA for sTNFR.
Levels of sTNFRSF1A were assayed using paired monoclonal antibodies (R&D Systems Europe) in a biotin–avidin sandwich ELISA system adapted from R&D Systems' Duoset kit design. Briefly, it consisted of coating the plates (Maxisorb; Nunc, Uxbridge, UK) with 2 μg/ml of anti-TNFRSF1A capture monoclonal antibody, and blocking the wells for 1 hour with 1% bovine serum albumin (BSA; Sigma), 5% sucrose, and 0.05% NaN3 in phosphate buffered saline (PBS). All incubations were performed at room temperature. After washing, the diluted samples (1:10 or 1:100, as appropriate, in reagent diluent consisting of 0.1% BSA, 0.05% Tween 20 in Tris buffered saline) or standard (recombinant human sTNFRSF1A; R&D Systems Europe) were added. After incubation for 2 hours and washing, biotinylated anti-TNFRSF1A monoclonal antibody (0.1 μg/ml) was added, and following a further 2-hour incubation and wash, ExtrAvidin (1:1,000 dilution; Sigma) was added, incubation continued for 20 minutes, and tetramethylbenzidine (Merck Biosciences, Nottingham, UK) substrate was added for 30 minutes. The reaction was stopped with 1M H2SO4, and the optical density (450 nm) of the substrate was measured on an Emax Microplate Reader (Molecular Devices, Wokingham, UK). Levels of sTNFRSF1B were assayed by Quantikine ELISA according to the manufacturer's instructions.
Surface expression of TNFRSF1A on fibroblasts.
The surface expression of TNFRSF1A was assessed by cell ELISA. The fibroblasts showed very high autofluorescence and therefore could not be analyzed by flow cytometry. The cell numbers were assessed by alamarBlue conversion as described above. The cells were washed with PBS and incubated with an unconjugated anti-TNFRSF1A monoclonal antibody (2 μg/ml; R&D Systems Europe) or PBS alone (as a conjugate-only control) for 1.5 hours at 4°C, then washed with cold PBS. An alkaline phosphatase–conjugated anti-mouse IgG antibody (1:1,000 dilution in PBS; Sigma) was then added and incubated as before. The cells were again washed with cold PBS, incubated for 30 minutes with p-nitrophenyl phosphate substrate (1 mg/ml; Sigma) dissolved in diethanolamine buffer, and the optical density was read at 405 nm.
Production of recombinant WT and mutant TNFRSF1A DNA.
The full-length TNFRSF1A coding sequence was amplified from complementary DNA by polymerase chain reaction using flanking primers designed from previously published sequences (GenBank accession no. NM_001065). DNA was amplified with Elongase (Invitrogen), cloned into the vector pcDNA4/TO (Invitrogen) and sequenced from vector and internal primary sites using BigDye 2 terminators on an ABI Prism 310 sequencer (both from Applied Biosystems, Warrington, UK). To remove the cytoplasmic signaling domain, a mutant (Δ-sig) was constructed using site-directed mutagenesis (Qwikchange kit; Stratagene, Amsterdam, The Netherlands), with a stop codon introduced at residue 215 (numbered from the start of the mature TNFRSF1A sequence), leaving a 10-residue cytoplasmic tail. Mutant receptors (C33Y, T50M, C52F, C88Y, and R92Q) were produced from the Δ-sig variant using site-directed mutagenesis. All products were sequenced in both directions along the full length of the TNFRSF1A coding region to ensure that only the desired mutation(s) had been introduced. Plasmid isolation from bulk cultures for transfection into eukaryotic cells was carried out using an Endofree Plasmid Maxiprep kit (Qiagen, Crawley, UK).
Transfection and cloning of cell lines expressing recombinant TNFRSF1A.
The tetracycline-regulated expression (T-Rex) 293 human embryonic kidney cell line was used for transfections. The 293 cell line has been reported to express low levels of TNF receptors, predominantly TNFRSF1A (24); however, in our experience, TNFRSF1A and TNFRSF1B expression by 293 cells was undetectable by flow cytometry (data not shown).
The 293 cells were stably transfected with the WT and single-mutation Δ-sig constructs using FuGENE 6 (Roche Diagnostics, Lewes, UK) according to the manufacturer's protocols. A 3:1 ratio of FuGENE 6 reagent to DNA was used. Ninety-six hours posttransfection, the T-Rex 293 cells were switched to selective medium containing 10% FCS in DMEM supplemented with 100 units/ml of penicillin, 100 μg/ml of streptomycin, 2 mML-glutamine, 10 mM HEPES buffer, 5 μg/ml of Blasticidin S HCl (Invitrogen), and 400 μg/ml of Zeocin (Invitrogen). Initial transfection success was assessed using pcDNA4/TO LacZ control plasmid expression and was detected with a β-galactosidase staining kit (Invitrogen).
Primary selection of transfected cells was for 21 days, after which the cells were induced to express receptor. Cultured cells were split 1:2, and 1 μg/ml of doxycycline (a tetracycline derivative; Sigma) was added. Twenty hours postinduction, the cells were assayed for surface-expressed TNFRSF1A by flow cytometry using phycoerythrin-labeled mouse anti-human TNFRSF1A (R&D Systems Europe) or phycoerythrin-labeled mouse IgG1 negative control (Dako, Ely, UK), and assayed on an EPICS XL flow cytometer (Beckman Coulter, High Wycombe, UK). Limiting dilution and subsequent selection of individual colonies produced cloned cell lines of the transfectants. Cloned populations of Δ-sig WT and mutant TNFRSF1A-containing cell lines were selected from doxycycline-induced cell clones that expressed high levels of receptor.
Production of dual transfectants expressing both recombinant mutant and WT TNFRSF1A.
The WT Δ-sig TNFRSF1A DNA produced as described above was cloned into the vector pDONR201 and subsequently into pDEST47 using the Gateway technology system (Invitrogen). The product was sequenced using BigDye 2 terminators on an ABI Prism 310 instrument to ensure correct sequence. Plasmids were isolated from bulk cultures for transfection into the previously transfected WT and mutant Δ-sig TNFRSF1A-containing cells described above. Transfection of the new construct was with FuGENE 6 at a 3:1 ratio of reagent to DNA, according to the manufacturer's protocol. Dual transfectants were selected with 400 mg/ml of Geneticin (Invitrogen) in addition to Blasticidin and Zeocin.
Induction and inhibition of TNFRSF1A shedding.
WT and mutant HEK 293 cell lines were induced to express TNFRSF1A as above. All the following procedures were performed on ice where possible. Cells were harvested and washed in DMEM containing 1% FCS, followed by centrifugation at 400g for 5 minutes. Cells were incubated in DMEM and 1% FCS with 10 ng/ml of PMA for 15 minutes at 37°C, with periodic mixing. In some experiments, duplicate tubes were established that also contained 30 mM TNFα protease inhibitor 1 (TAPI-1; Merck Biosciences). Cells that were not treated with PMA were kept on ice. Cells were washed in DMEM (1% FCS) then in PBS–0.5% BSA, and resuspended in PBS–0.5% BSA. Cells were mixed with mouse phycoerythrin-labeled anti-human TNFRSF1A antibody or mouse IgG1 (negative control) and incubated on ice for 30 minutes in the dark. Cells were subsequently washed twice with PBS–0.5% BSA, resuspended in 0.5% formaldehyde fixative, and analyzed by flow cytometry.
Analysis of TNFα binding.
WT and mutant Δ-sig TNFRSF1A cell lines were induced with doxycycline to express TNFRSF1A, as described above. Recombinant human TNFα was added at a final concentration of 50 ng/ml, and the cells were incubated at 37°C for 2 hours. Cells were harvested and washed in PBS–0.5% BSA and stained with primary antibodies (mouse anti-human TNFα; BioSource International, Fleurus, Belgium) or mouse IgG2a negative control (Dako) for 30 minutes on ice. Cells were washed twice with PBS–0.5% BSA and mixed with fluorescein isothiocyanate–labeled goat anti-mouse Ig F(ab′)2 (Dako) for 30 minutes on ice. Cells were washed twice, resuspended in 0.5% formaldehyde fixative, and analyzed by flow cytometry. Duplicate cells were stained for surface expression of TNFRSF1A as above to confirm doxycycline induction. Controls showed very little nonspecific staining of the cells by anti-human TNFα if TNFα was not added or if TNFα was added but TNFRSF1A expression was not induced with doxycycline.
Analysis of the data from the sTNFRSF1A fibroblast- and leukocyte-shedding experiments was performed using Student's t-test with 95% confidence intervals. The data are presented as the mean ± SEM.
TNFRSF1A receptor shedding from leukocytes in response to PMA.
Stimulation of whole blood cells with 100 ng/ml of PMA demonstrated that there was no significant difference in the shedding of either TNFRSF1A or TNFRSF1B from the surface of leukocytes obtained from patients with the C33Y mutation compared with those from normal controls, as determined by a capture ELISA system to detect sTNFRSF1A or sTNFRSF1B (Figures 1A and B). Flow cytometric analysis of neutrophil and monocyte surface TNFRSF1A also did not show a clear defect in the loss of receptors on cells from the patients as compared with those from normal controls in response to PMA (data not shown).
TNFRSF1A receptor shedding from fibroblasts.
Shedding of sTNFRSF1A from fibroblasts in response to PMA was significantly lower for the patients than for the normal controls (P < 0.05 for both patient–control pairs at 48 hours) (Figure 2A). A small difference that approached statistical significance was also seen for spontaneous shedding from the untreated cells (P = 0.06 for both patient–control pairs at 48 hours) (Figure 2B). The results shown in Figures 2A and B are the mean of 5 separate experiments, with good experimental reproducibility. Alamar blue conversion indicated that the cell viability and density were similar for patient and control fibroblasts in each pairing. Therefore, the difference in TNFRSF1A shedding cannot be explained by there being fewer cells in the patient fibroblast cultures. No TNFRSF1B shedding defect was seen (data not shown).
Surface expression of TNFRSF1A on fibroblasts.
To exclude the possibility that the lower shedding of TNFRSF1A by fibroblasts from the TRAPS patients was due to lower than normal expression of TNFRSF1A, the surface expression of TNFRSF1A was assessed using a cell ELISA. The results were expressed as a ratio of the optical densities in the ELISA relative to the alamar blue conversion as a measure of cell numbers and viability in the cultures, and are reported as the mean of duplicate wells. The assays were performed on different occasions for each patient–control pairing, and are thus presented as separate graphs (Figures 3A and B). Surface expression of TNFRSF1A on cells from the patients was very similar to that on cells from the corresponding normal controls, indicating that the lower level of sTNFRSF1A shed by fibroblasts from the patients was not due to a substantially reduced surface expression of TNFRSF1A on their fibroblasts. Mutant or WT receptors could not be distinguished.
Surface expression and shedding of Δ-sig WT and mutant TNFRSF1A.
We produced WT and mutant recombinant clones of the TNFRSF1A gene truncated at residue 214 because, as mentioned previously, transfection with full-length TNFRSF1A induces apoptosis. The stably transfected HEK 293 cell lines, with the plasmids containing the truncated Δ-sig WT or mutant TNFRSF1A genes, were treated with doxycycline to induce expression of the recombinant TNFRSF1A. Cell surface TNFRSF1A was detected on all the cell lines by flow cytometry following staining with phycoerythrin-labeled anti-TNFRSF1A monoclonal antibody; clones showing high surface expression were derived from the lines (Figure 4A).
Substantial shedding of the surface-expressed TNFRSF1A was induced in the WT and in each of the mutant cell clones by treatment of the doxycycline-induced cells with PMA (Figure 4A). The cell lines expressing the C33Y and C52F mutant receptors showed very similar levels of surface TNFRSF1A induction; however, the shedding of the C52F receptor was reproducibly slightly less than that of the C33Y receptor (Figure 4A). Also, shedding of the R92Q mutant receptor was reproducibly slightly less than that of the WT TNFRSF1A, even though the levels of receptor induction on these 2 cell lines were virtually identical (Figure 4A). The T50M mutant cell line showed bimodal expression of surface TNFRSF1A prior to PMA stimulation, but showed substantial shedding of receptor after stimulation. However, apart from the minor variations noted above, no major differences in the degree of TNFRSF1A shedding were observed between the cell lines expressing WT or mutant forms of the receptor.
It has previously been shown that the metalloprotease inhibitor TAPI-1 blocks PMA-induced shedding of TNFRSF1A (19). Indeed, the PMA-induced loss of cell surface TNFRSF1A was substantially reduced in the presence of TAPI-1 on the WT and on all the mutant cell lines (Figure 4B). This is consistent with PMA-activated metalloprotease cleavage being responsible for the shedding of both WT and mutant receptors.
Shedding of coexpressed Δ-sig WT and mutant TNFRSF1A.
The HEK 293 cell line is reported to express naturally very low levels of TNF receptors (mainly TNFRSF1A) (24). Thus, in the transfected cell lines used in the experiments described above (single transfectants), the majority of TNFRSF1A expressed following treatment with doxycycline will be that encoded by the highly expressed transgene (either WT or mutant alone). However, it is likely that the patients' cells coexpress WT and mutant TNFRSF1A at similar levels (although this cannot be formally proven without antibodies that distinguish the WT and mutant types). We therefore produced cell lines cotransfected with the recombinant genes encoding mutant and WT Δ-sig TNFRSF1A (dual transfectants) in order to assess receptor shedding under the circumstances of coexpression of WT and mutant receptors. These dual transfectants were produced by retransfecting the original single transfectants (expressing either WT or mutant Δ-sig TNFRSF1A) with the recombinant WT Δ-sig TNFRSF1A gene in a plasmid vector carrying a different selection marker. Transfected cells were then selected under conditions requiring the presence of both plasmids (see Materials and Methods).
In order to confirm the expression of WT TNFRSF1A by the newly introduced plasmid, the TNFα binding capacity of the dual transfectants was examined. We have previously observed that the WT Δ-sig TNFRSF1A single-transfectant HEK 293 cell line showed good binding of TNFα, whereas the T50M Δ-sig single transfectant showed much lower binding, and the C33Y and C52F Δ-sig single transfectants showed no TNFα binding despite induced surface expression of TNFRSF1A (Todd et al: submitted for publication). Interestingly, the R92Q Δ-sig single transfectant showed TNFα binding similar to that of the WT (Todd I, et al: unpublished observations). As shown in Figure 5A, these findings were confirmed under conditions in which all the cell lines showed surface expression of their transfected TNFRSF1A. In contrast, the results shown in Figure 5B demonstrate that TNFα was bound by all the dual transfectants expressing surface Δ-sig TNFRSF1A, thus indicating coexpression of the newly introduced WT Δ-sig TNFRSF1A. This also demonstrates that WT TNFRSF1A can still bind TNFα in the presence of the mutant receptors.
The single-transfectant cell lines (Figure 6A) and the dual-transfectant cell lines (Figure 6B) demonstrated very similar shedding of TNFRSF1A upon treatment with PMA. Thus, similar TNFRSF1A shedding occurs with mutant receptor alone as with mutant plus WT receptors, and was induced to a substantial extent in all cases.
Cell lines transfected with an additional TNFRSF1A mutant (C88Y) were included in the experiments shown in Figures 5 and 6. Like R92Q, C88Y is a mutation within CRD2, whereas the other mutations examined (C33Y, C52F, and T50M) are within CRD1. Although the level of C88Y expression achieved was relatively low, substantial receptor shedding was seen upon treatment with PMA (Figure 6). Like the other cysteine mutants (but not the R92Q mutant), the C88Y mutant receptor did not bind TNFα (Figure 5A).
Not all TRAPS-related TNFRSF1A mutations result in reduced receptor shedding by leukocytes (4, 14), but there is evidence that serum TNFRSF1A is low in patients with TRAPS (1, 20). However, no previous studies have investigated other types of cells from TRAPS patients. In addition, no previous study has assessed the shedding from homozygous cells expressing high levels of only TRAPS-related mutant TNFRSF1A. Here we demonstrated that the defective receptor shedding varies not only between mutations, but also between cell types bearing the same mutation, and is not purely a function of the mutations in TNFRSF1A.
Despite an earlier finding that the C52F mutation was associated with reduced shedding from leukocytes (1), we have shown here that leukocytes from C33Y TRAPS patients do not show any defect in PMA-induced TNFRSF1A shedding. However, dermal fibroblasts from the same C33Y TRAPS patients had a marked reduction in TNFRSF1A shedding. This difference between cell types indicates that factors in addition to the mutations in TNFRSF1A influence receptor shedding. It also suggests that there is a difference between the two cell types in their roles in the inflammatory response seen in these patients. The observed difference is not dependent on the dose of PMA used (10 ng/ml with fibroblasts and 100 ng/ml with leukocytes), since treating leukocytes with 10 ng/ml also failed to distinguish TNFRSF1A shedding between C33Y TRAPS patients and controls (data not shown). The method used to induce TNFRSF1A shedding from the leukocytes was identical to that used for the other TRAPS families, such as patients with C52F TRAPS, in which a reduction in the shedding of TNFRSF1A was observed.
The method chosen to detect TNFRSF1A shedding from the fibroblasts differed slightly from the method used for the leukocytes. In particular, fibroblasts must adhere to the culture plate to remain viable, resulting in 100-fold fewer cells for analysis than was available for the leukocytes. A longer incubation period was therefore necessary for the sTNFRSF1A released by these cells to be within the detection range of the ELISAs used. However, further samples taken at 72 and 96 hours after PMA stimulation showed a continued release of sTNFRSF1A from the normal fibroblasts, suggesting that the time subsequent to stimulation did not limit the release of sTNFRSF1A from these cells (data not shown).
TRAPS is characterized by an augmented inflammatory response, and flares have been related to infection and trauma (17). Such inflammatory stimuli liberate proinflammatory cytokines (TNF, interleukin-1, and interleukin-6) and in TRAPS, defective negative feedback (such as reduced sTNFRSF1A production) may be unable to terminate these reactions. In the skin, there is a tightly regulated microenvironment of cytokines and growth factors produced by keratinocytes and dermal fibroblasts. Hence, fibroblasts in the skin and other tissues may also contribute to the feedback mechanisms. In this study, we have demonstrated that C33Y TRAPS dermal fibroblasts have a reduced ability to produce sTNFRSF1A. These results, together with the reduced serum levels of TNFRSF1A in these patients, suggest that fibroblasts may normally be an important source of sTNFRSF1A in the control of inflammation.
All of the HEK 293 Δ-sig cell lines expressing mutant forms of TNFRSF1A, as well as the WT, showed PMA-induced receptor shedding, although to differing extents. Since these recombinant receptors lacked the intracellular region, this finding is consistent with that of a previous study reporting that the intracellular region of TNFRSF1A is not essential for receptor shedding (25). In all cases, shedding was at least partially inhibited by the metalloprotease inhibitor TAPI-1, indicating that the mechanism of shedding of these recombinant truncated receptors is the same as the mechanism for natural TNFRSF1A (19).
Similar levels of shedding were seen in transfectants expressing high levels of mutant TNFRSF1A alone or mutant plus WT TNFRSF1A, suggesting that shedding behavior is similar in either situation. We also observed that most of the mutant forms of Δ-sig TNFRSF1A showed substantially reduced TNF binding but that, in the dual transfectants, WT TNFRSF1A still bound TNF in the presence of the mutant receptors. This is consistent with the previous report that leukocytes from TRAPS patients and healthy controls show similar affinities of binding for TNF (1).
Although minor differences in the degree of shedding from the transfectants were seen between WT and the various mutants of TNFRSF1A, all showed substantial shedding. All features of the transfected HEK 293 cell lines are very similar, apart from the mutations in the transfected TNFRSF1A. This, again, therefore indicates that potential shedding differences are not purely a function of TNFRSF1A structure, but are influenced by other genetic and/or cellular variables. This conclusion is consistent with the differences in C33Y TNFRSF1A shedding by leukocytes and fibroblasts, as discussed above. It is also consistent with the findings of Aganna et al (14), who reported that not only do TRAPS patients differ with respect to a TNFRSF1A shedding defect, but some patients with autosomal-dominant periodic fever without TNFRSF1A mutations also demonstrate reduced shedding of TNFRSF1A and have low serum levels of the soluble receptor.
It has been difficult to conceive how diverse mutations in the cysteine-rich domains in the distal regions of the TNFRSF1A ectodomain could influence shedding that is a result of cleavage proximal to the plasma membrane (26). However, it has recently been found that cleavage depends on interaction of the TNFRSF1A ectodomain with a type II integral membrane protein called aminopeptidase regulator of TNFRSF1A shedding (ARTS-1) (27). Although the precise sites of interaction between ARTS-1 and TNFRSF1A have not been defined, it is possible that TRAPS-associated mutations may reduce interactions between TNFRSF1A and ARTS-1 and therefore reduce shedding.
Another possibility that should be kept in mind is that mutations may affect the internalization of TNFRSF1A into the cytoplasm that has been reported to occur following TNF binding (28). This might, for example, prolong signaling, with inflammatory consequences.
In general, our data support the concept that a combination of functional abnormalities resulting from mutations in TNFRSF1A may contribute to TRAPS. Shedding defects are influenced not only by different mutations in TNFRSF1A, but also by other factors, including cell type. This raises the possibility that reduced shedding of TNFRSF1A may be one of several factors that determine the clinical picture of TRAPS and that other mechanisms of TNF receptor signaling and control of inflammation are also involved. The complex and variable clinical picture in TRAPS may relate to differences in several functional effects of the different mutations.
We are grateful to Dr. Jerôme Galon and Dr. Daniel Kastner for their assistance in the leukocyte shedding experiments, and to Dr. Elizabeth McDermott and Dr. Elizabeth Drewe for obtaining the samples from the patients.