Distribution of TNFA haplotypes in healthy CaucasiansComment on the articles by Newton et al and Zeggini et al.код для вставкиСкачать
ARTHRITIS & RHEUMATISM Vol. 50, No. 6, June 2004, pp 2034-2040 © 2004, American College of Rheumatology group of genotyped individuals, without the need for family data (10). Although the EM algorithm aims at finding haplotype frequencies that yield a maximum likelihood given the input genotypes, the procedure is such that alternative haplotype distributions with high likelihood estimations may exist in addition to the reported one. Thus, one can speculate that the discussed discrepancies are caused by different results (but still consistent with the data) of the procedure estimating the haplotypic frequencies. However, one fact argues against this possibility. Without performing haplotype estimations, our analysis confirmed that ⫺1031C and ⫺863A are in very strong (virtually complete) linkage disequilibrium (8). The strong association between ⫺863A and ⫺1031C (but not ⫺1031T) could be detected in our data by simple 2 ⫻ 2 contingency tables (P ⬍ 10⫺10), i.e., an approach free from the errors that are possible with the EM algorithm. In accordance with this notion, Zeggini et al (2) showed that the ⫺863A allele is found exclusively on haplotypes encoding ⫺1031C (Table 1). Conversely, Newton et al (1) did not find such a haplotype but instead reported a relatively high frequency of the haplotype positive for ⫺1031T and ⫺863A (Table 1). Thus, the discrepancies with the results of Newton et al (1) cannot be caused just by different solutions to the problem of the haplotypic associations. Whereas the distribution of TNF␣ promoter singlenucleotide polymorphisms has been shown to vary considerably in different ethnic groups (11), it is interesting that similar frequencies of TNF haplotypes were found in an Anglo-Saxon (UK) (2) and a Slavic (Polish) population. At the same time, it is not clear why considerably different results were obtained in the Caucasian UK population studied by Newton et al (1). DOI 10.1002/art.20283 Distribution of TNFA haplotypes in healthy Caucasians: comment on the articles by Newton et al and Zeggini et al To the Editor: We read with interest the articles by Newton et al (1) and Zeggini et al (2), who found evidence that a gene(s) located in the region of the tumor necrosis factor (TNF) locus may contribute to the susceptibility to rheumatoid arthritis (RA) and juvenile idiopathic arthritis (JIA), respectively. Studies analyzing the association of non-HLA genes encoded in the major histocompatibility complex (MHC) with susceptibility to HLA-linked diseases such as RA and JIA are important; however, they are complicated by the relatively large number of potentially relevant polymorphisms as well as the complex patterns of linkage disequilibrium occurring in the MHC region (3,4). A way to approach this problem, which was adopted by both groups of investigators, relies on the analysis of haplotypes rather than individual markers. The analysis of haplotypes may reveal biologically relevant associations that are not apparent on analysis of individual markers. Further, analysis of large numbers of individual markers can be considerably simplified when the haplotypic associations among them are elucidated (5). In accordance with this notion, both studies showed that the majority of genetic variation in the TNF locus can be explained by 4 (1) or 5 (2) haplotypes with frequencies of ⬎5% in the healthy population. However, it is surprising that although both studies were performed in Caucasians living in the UK there are considerable differences between the frequencies of TNF haplotypes reported as representative of a normal population. For example, the TNF haplotype defined by the presence of nucleotides ⫺1031T, ⫺863C, ⫺308G, 851A, and 1304A (numbering of positions relative to the start of transcription) had a frequency of 56% in the population studied by Zeggini et al but only 5% in the population investigated by Newton et al (2 ⫽ 108, 1df, P ⬍ 10⫺9). A number of other differences between the results of the 2 studies (1,2) are shown in Table 1. To further study the frequencies of TNF haplotypes in Caucasians, we analyzed the distribution of 5 polymorphisms in the promoter region of the TNF␣ gene (⫺1031T/C, ⫺863C/A, ⫺857C/T, ⫺308G/A, ⫺238G/A) among 248 healthy, unrelated Polish adults from previously described cohorts (6,7). Polymorphisms in the 5⬘ flanking region of the TNF␣ gene were identified by dot-blot hybridization with sequencespecific oligonucleotide probes, as previously described (8), and the haplotypes were derived using expectationmaximization (EM) algorithm implemented using the Arlequin software package (9). As can be seen in Table 1, we found evidence for the presence in a Polish population of 5 TNFA haplotypes with frequencies exceeding 5%, which together accounted for 99% of all haplotypes. The frequencies of individual haplotypes were similar to those reported by Zeggini et al (2) but were considerably different from those found by Newton et al (1) (Table 1). The EM algorithm implemented with the Arlequin software package allows, in a typical situation, an estimation with ⬎95% accuracy of the distribution of haplotypes in a Rafal Ploski, MD, PhD Warsaw Medical University Warsaw, Poland Tomasz Bednarczuk, MD, PhD Polish Academy of Science Warsaw, Poland Yuji Hiromatsu, MD, PhD Kurume University School of Medicine Kurume, Japan 1. Newton J, Brown MA, Milicic A, Ackerman H, Darke C, Wilson JN, et al. The effect of HLA–DR on susceptibility to rheumatoid arthritis is influenced by the associated lymphotoxin ␣–tumor necrosis factor haplotype. Arthritis Rheum 2003;48:90–6. 2. Zeggini E, Thomson W, Kwiatkowski D, Richardson A, Ollier W, Donn R. Linkage and association studies of single-nucleotide polymorphism–tagged tumor necrosis factor haplotypes in juvenile oligoarthritis. Arthritis Rheum 2002;46:3304–11. 3. Smerdel A, Lie BA, Ploski R, Koeleman BP, Førre O, Thorsby E, et al. A gene in the telomeric HLA complex distinct from HLA–A is involved in predisposition to juvenile idiopathic arthritis. Arthritis Rheum 2002;46:1614–9. 4. Ploski R, Vinje O, Rønningen KS, Spurkland A, Sørskaar D, Vartdal F, et al. HLA class II alleles and heterogeneity of juvenile rheumatoid arthritis: DRB1ⴱ0101 may define a novel subset of the disease. Arthritis Rheum 1993;36:465–72. 5. Schork NJ, Fallin D, Lanchbury JS. Single-nucleotide polymorphisms and the future of genetic epidemiology. Clin Genet 2000; 58:250–64. 6. Bednarczuk T, Hiromatsu Y, Fukutani T, Jazdzewski K, Miskiewicz P, Osikowska M, et al. Association of Cytotoxic T-lymphocyte-associated antigen-4 (CTLA-4) gene polymorphism 2034 LETTERS 2035 Table 1. Comparison of distribution of TNFA haplotypes among healthy Polish subjects with data reported as being representative of UK Caucasian controls* Population (ref.) TNFA haplotype ⫺1031 T T T T C C C C T Sum ⫺863 C C C C A C C C A ⫺857 C T C T C C C ⫺308 G G A A G G G A G ⫺238 G G G G G G A Polish Caucasians (n ⫽ 496) UK Caucasians (2) (n ⫽ 162) 51 ⫾ 2 16 ⫾ 2 16 ⫾ 2 0 12 ⫾ 1 0 5⫾1 0 0 99 UK Caucasians (1) (n ⫽ 200) 51 12 13 2 11 6 4 NR NR 99 P1† P2‡ 8 ⫺6 ⬍10 ⬍10⫺6 2 ⬍10⫺5 ⬍10⫺3 NR 55 ⬍10⫺4 ⬍10⫺6 ⬍10⫺3 ⬍10⫺6 15 15 95 ⬍10⫺6 ⬍10⫺6 ⬍10⫺4 ⬍10⫺4 * Values are mean ⫾ SD percentage or the percentage. Positions ⫺857 and ⫺238 were not analyzed by Newton et al (1). There were no statistically significant differences in the distribution of these haplotypes between Polish Caucasians and the UK Caucasians studied by Zeggini et al (2). TNF ⫽ tumor necrosis factor; n ⫽ number of chromosomes; NR ⫽ not reported. † UK Caucasians, Newton et al versus Polish Caucasians. ‡ UK Caucasians, Newton et al versus UK Caucasians, Zeggini et al. 7. 8. 9. 10. 11. and non-genetic factors with Graves’ ophthalmopathy in European and Japanese populations. Eur J Endocrinol 2003;148:13–8. Ploski R, Ronningen KS, Thorsby E. HLA class II profile of a Polish population: frequencies of DRB1, DQA1, DQB1, and DPB1 alleles and DRB1-DQA1-DQB1 haplotypes. Transplant Proc 1996;28:3431–2. Higuchi T, Seki N, Kamizono S, Yamada A, Kimura A, Kato H, et al. Polymorphism of the 5⬘-flanking region of the human tumor necrosis factor (TNF)-␣ gene in Japanese. Tissue Antigens 1998; 51:605–12. Schneider S, Roessli D, Excoffier L. Arlequin: a software for population genetics data analysis: version 2.000. Geneva: Genetics and Biometry Lab, Department of Anthropology, University of Geneva; 2000. Fallin D, Schork NJ. Accuracy of haplotype frequency estimation for biallelic loci, via the expectation-maximization algorithm for unphased diploid genotype data. Am J Hum Genet 2000;67: 947–59. Baena A, Leung JY, Sullivan AD, Landires I, Vasquez-Luna N, Quinones-Berrocal J, et al. TNF-␣ promoter single nucleotide polymorphisms are markers of human ancestry. Genes Immun 2002;3:482–7. DOI 10.1002/art.20455 Reply To the Editor: We thank Ploski and colleagues for their interest in our study. The explanation for the difference in our findings is a typographic error in Table 2 of our article, whereby the alleles for marker TNF ⫺1031 were labeled incorrectly. The correct frequencies are shown here in Table 1. Using the correct allele frequencies, there is little difference between the 2 studies of UK Caucasians (P ⫽ 0.03) (1), but quite marked differences are apparent when comparing either of the UK studies with the Polish study (for Newton et al, P ⫽ 7 ⫻ 10⫺9; for Zeggini et al, P ⫽ 5 ⫻ 10⫺7). The 2 probable explanations for the difference between the findings of Ploski et al and those of the 2 UK studies are that the UK studies were family studies, whereas the Polish study involved only unrelated individuals, and ethnic differences in haplotype frequencies. Family-based methods are more accurate in predicting haplotypes than are methods in which unrelated individuals are used, but it is reassuring that for common haplotypes, relatively little difference was observed using the different methods. Our study used both the PHAMILY and PHASE (2) programs to determine haplotypes, whereas Zeggini et al used the GeneHunter program (3). We also analyzed our data using TRANSMIT (4) and found no significant difference between Table 1. TNFA haplotype frequencies in British and Polish subjects* Population, author TNFA haplotype Polish UK UK Caucasians, Caucasians, Caucasians, Plotski Zeggini Newton et al et al et al ⫺1031 ⫺863 ⫺857 ⫺308 ⫺238 (n ⫽ 496) (n ⫽ 162) (n ⫽ 200) T T T T C C C C T C C C C A C C C A C T C T C C C G G A A G G G A G G G G G G G A 51 16 16 0 12 0 5 0 0 51 12 13 1 11 6 4 NR NR 54 4 15 0 15 4.4 1 0 0 * Values are the percentage. n ⫽ number of chromosomes; NR ⫽ not reported. 2036 LETTERS the results obtained with TRANSMIT and the PHAMILY/ PHASE results. In situations such as that of the MHC where there is strong linkage disequilibrium, haplotyping programs such as the EM algorithm or PHASE are likely to perform well in unrelated individuals. EM algorithm–based programs such as Arlequin (5) are attractively simple to use. However, because they do not report posterior haplotype probabilities, they cannot be generally recommended, because in the absence of significant linkage disequilibrium the predictions of all haplotyping programs are insecure. In this setting, Bayesian programs such as PHASE have the major advantage that, unlike Arlequin and other EM algorithm programs, the accuracy of the haplotype prediction is reported. Julia Newton, MRCP Dominic Kwiatkowski, FRCP Wellcome Trust Centre for Human Genetics Oxford, UK Paul Wordsworth, FRCP Matthew A. Brown, MD, FRACP University of Oxford Oxford, UK 1. Zeggini E, Thomson W, Kwiatkowski D, Richardson A, Ollier W, Donn R. Linkage and association studies of single-nucleotide polymorphism–tagged tumor necrosis factor haplotypes in juvenile oligoarthritis. Arthritis Rheum 2002;46:3304–11. 2. Stephens M, Smith NJ, Donnelly P. A new statistical method for haplotype reconstruction from population data. Am J Hum Genet 2001;68:978–89. 3. Pratt SC, Daly MJ, Kruglyak L. Exact multipoint quantitative-trait linkage analysis in pedigrees by variance components. Am J Hum Genet 2000;66:1153–7. 4. Clayton D. A generalization of the transmission/disequilibrium test for uncertain-haplotype transmission. Am J Hum Genet 1999;65: 1170–7. 5. Schneider S, Roessli D, Excoffier L. Arlequin: a software for population genetics data analysis: version 2.000. Geneva: Genetics and Biometry Lab, Department of Anthropology, University of Geneva; 2000. DOI 10.1002/art.20282 Bone mass in adolescents with early-onset juvenile idiopathic arthritis: comment on the article by Lien et al To the Editor: The issue of skeletal integrity and achievement of peak bone mass is crucial in a chronic disease such as juvenile idiopathic arthritis, as noted by Lien et al in their recent report (1). It is therefore very important to conduct long-term outcome studies as was done by Lien and colleagues. In addition, in a very detailed examination protocol, they addressed many factors that might influence skeletal development. The conclusions drawn are nevertheless hampered by the fact that height and bone geometry have a strong effect on bone mineral density (BMD) and bone mineral content (BMC) measured by dual x-ray absorptiometry (DXA). The authors refer to this problem in their Patients and Methods section, where they state that in children, total-body measurements are preferable and BMC is preferable to BMD, but both parameters will still be reduced in a growth-retarded child whose BMD and bone geometry might be perfectly normal for body size. The adolescents with low total-body BMC in their study were significantly smaller and weighed less than those with normal total-body BMC. The only other parameters with significant differences at the same P value (⬍0.001) were lean mass, physician’s global assessment, number of actively involved joints, and number of erosions (Table 3 in ref. 1). It is not stated whether the lower height and weight represented reduced values as compared with age-matched reference data, but the impression from the data presented is that the adolescents with low BMC had a longer and more severe disease course, less physical activity, were smaller, and had less muscle (lean mass). One could therefore conclude that low BMD and BMC are explained at least in part by less bone due to smaller body size compared with age-matched controls. The skeleton of these patients might actually be perfectly normal for their body size. The conclusion of Lien et al that the future fracture risk will be increased cannot be drawn from these data. If these patients remain smaller than normal but their bones are adapted to their body size and muscle mass, there is not necessarily an increased fracture risk. The interaction of muscle and bone is not addressed in the Discussion section of Lien and colleagues’ report. Muscle mass/muscle force is the single most important factor influencing bone growth and maintenance (2), and in healthy children, adolescents, and adults, muscle mass and bone mass are strongly correlated (3). The reduced bone mass in the study by Lien et al might actually be the result of the reduced lean mass which, in turn is very well explained by body size, longer disease duration, and a higher number of actively involved joints, which might lead to inactivity. Whether there are true abnormalities in bone mass in these patients cannot be determined without proper correction for body/bone size (4), and whether there are true abnormalities in cortical density or cortical thickness cannot be determined by DXA. The only method to answer this question would be a true 3-dimensional measurement with quantitative computed tomography (5). Johannes Roth, MD Charité Virchow Klinikum Berlin, Germany 1. Lien G, Flato B, Haugen M, Vinje O, Sorskaar D, Dale K, et al. Frequency of osteopenia in adolescents with early-onset juvenile idiopathic arthritis: a long-term outcome study of one hundred five patients. Arthritis Rheum 2003;48:2214–23. 2. Frost HM, Schoenau E. The “muscle bone unit” in children and adolescents: a 2000 overview. J Pediatr Endocrinol Metab 2000; 13:571–90. 3. Schiessl H, Frost HM, Jee WS. Estrogen and bone-muscle strength and mass relationships. Bone 1998;22:1–6. 4. Molgaard C, Thomsen BL, Prentice A, Cole TJ, Michaelsen KF. Whole body bone mineral content in healthy children and adolescents. Arch Dis Child 1997;76:9–15. 5. Schoenau E. Problems of bone analysis in childhood. Pediatr Nephrol 1998;12:420–9. LETTERS 2037 DOI 10.1002/art.20281 Concerns in reporting of serum vascular endothelial growth factor levels: comment on the article by Nakahara et al To the Editor: I read with great interest the article by Nakahara et al (1) reporting a reduction in serum levels of vascular endothelial growth factor (VEGF) after anti–interleukin-6 receptor monoclonal antibody (anti–IL-6R mAb) therapy in patients with rheumatoid arthritis. I would like to call attention to some methodologic concerns. When VEGF is measured, the conditions of processing (e.g., length of time and force of centrifugation, interval between sample collection and processing) are relevant; they should be standardized and specified. First, concentrations of VEGF measured in serum samples may reflect blood platelet degranulation in vitro rather than synthesis of VEGF by peripheral tissues (2–4). In plasma, platelet degranulation is minimized by adding anticoagulants to the blood samples, and as a consequence, plasma VEGF concentrations are up to 20 times lower than matched serum VEGF concentrations (4). Citrated, EDTA-treated, or heparinized plasma processed in glass tubes is the material of choice for measurement of circulating extracellular VEGF. Second, the serum VEGF concentration changes according to clotting duration (2,3). Hormbrey et al found that there is no significant difference in VEGF levels in serum samples processed between 30–60 minutes after collection versus immediate processing, but the difference becomes significant when the samples are processed after 2–6 hours (3). In a clinical situation, where blood samples are kept for variable amounts of time before processing, the contribution of the clotting processing may interfere with the measurement of circulating VEGF levels at the time of sampling. Third, even if a uniform clotting time could be applied to all samples, between-subject variation in generation of VEGF in clotted samples may make the interpretation of any observed difference between disease and control groups very difficult and may invalidate the results (4). Finally, when using serum for VEGF measurement, it may be advisable to correct the measurements to platelet count (5). In conclusion, it would be of interest to confirm the effect of anti–IL-6R mAb therapy on extracellular circulating VEGF levels by performing the measurements on plasma samples. Simone Ferrero, MD San Martino Hospital University of Genoa Genoa, Italy 1. Nakahara H, Song J, Sugimoto M, Hagihara K, Kishimoto T, Yoshizaki K, et al. Anti–interleukin-6 receptor antibody therapy reduces vascular endothelial growth factor production in rheumatoid arthritis. Arthritis Rheum 2003;48:1521–9. 2. Webb NJ, Bottomley MJ, Watson CJ, Brenchley PE. Vascular endothelial growth factor (VEGF) is released from platelets during blood clotting: implications for measurement of circulating VEGF levels in clinical disease. Clin Sci 1998;94:395–404. 3. Hormbrey E, Gillespie P, Turner K, Han C, Roberts A, McGrouther D, et al. A critical review of vascular endothelial growth factor (VEGF) analysis in peripheral blood: is the current literature meaningful? Clin Exp Metastasis 2002;19:651–63. 4. Banks RE, Forbes MA, Kinsey SE, Stanley A, Ingham E, Walters C, et al. Release of the angiogenic cytokine vascular endothelial growth factor (VEGF) from platelets: significance for VEGF measurements and cancer biology. Br J Cancer 1988;77:956–64. 5. Verheul HM, Hoekman K, Luykx-de Bakker S, Eekman CA, Folman CC, Broxterman HJ, et al. Platelet: transporter of vascular endothelial growth factor. Clin Cancer Res 1997;3:2187–90. DOI 10.1002/art.20305 Intraarticular injection of anti–tumor necrosis factor: comment on the letter by Arnold et al To the Editor: The report by Arnold et al of a case of a presumed adverse reaction to etanercept injected into the affected knee of a patient with psoriatic arthritis (1) may, as the authors suggest, be an example of a local immune/allergic reaction to etanercept, by analogy to the injection site reactions seen fairly frequently during the administration of etanercept by the usual subcutaneous route. Arnold and colleagues’ patient received high-dose antibiotics immediately after the reaction occurred. In spite of negative findings on cultures of fluid later drawn from the joint, there may have been an infection that was resolved with the antibiotic treatment. There is a risk, albeit small, of low-grade subclinical infection at the time of a second injection, and it would seem appropriate to wait a reasonable interval of time between injections. In a series of 25 patients who received 1 injection of etanercept, my colleagues and I did not observe any troublesome reactions (2). More than 10 of our patients have received further injections of etanercept in the same or other joints, with no local reaction. However, in all cases, there was an interval of ⬎3 months between the first and second injections. In some of our patients, smaller joints were injected, with varying effect, including some cases in which the treatment was highly successful (3). This is in contrast to the negative results reported by Arnold et al (1). Similar conflicting results have been observed with infliximab treatment (4,5). It is likely that a number of factors influence the effect, including the type of arthritis and the joint treated, and previous lack of response to steroid injection may be a predictor of lack of response to anti–tumor necrosis factor injection. Finally, injections should be administered with ultrasound guidance for positioning, because even after knee puncture with fluid aspiration, it is possible that the subsequent injection may be inadvertently administered outside the joint (6). This can be easily avoided with ultrasonographic monitoring (7). For scientific assessment of the value of anti–tumor necrosis factor injection into joints, investigators must be assured that the treatment indeed reached the intended site. Henning Bliddal, MD Parker Institute Frederiksberg, Denmark 2038 1. Arnold EL, Khanna D, Paulus H, Goodman MP. Acute injection site reaction to intraarticular etanercept administration [letter]. Arthritis Rheum 2003;48:2078–9. 2. Bliddal H, Terslev L, Qvistgaard E, Kristoffersen H, Torp-Pedersen S, Danneskiold-Samsoe B. Injection of etanercept into arthritic joints. I. Safety. Ann Rheum Dis 2002; 61 Suppl 1:169. 3. Bliddal H, Qvistgaard E, Terslev L, Savnik A, Holm CC, Danneskiold-Samsoe B, et al. Injection of etanercept into arthritis joints: dose-response and efficacy [abstract]. Arthritis Rheum 2002;46 Suppl 9:S518. 4. Bokarewa M, Tarkowski A. Local infusion of infliximab for the treatment of acute joint inflammation. Ann Rheum Dis 2003;62: 783–4. 5. Kellner H, Kroetz M, Schattenkirchner M, Kellner W. Successful therapy of sacroileitis in AS patients by intraarticular injection of infliximab [abstract]. Arthritis Rheum 2002;46 Suppl 9:S431. 6. Jones A, Regan M, Ledingham J, Pattrick M, Manhire A, Doherty M. Importance of placement of intra-articular steroid injections. BMJ 1993;307:1329–30. 7. Bliddal H, Torp-Pedersen S. Use of small amounts of ultrasound guided air for injections. Ann Rheum Dis 2000;59:926–7. DOI 10.1002/art.20456 LETTERS Figure 1. Interleukin-1␤ (IL-1␤)–induced expression of aggrecanase 1 (ADAMTS-4) RNA and protein in bovine chondrocytes and human synovial fibroblasts. A, Analysis of untreated (control) and IL-1– treated bovine chondrocyte RNA by Northern hybridization with ADAMTS-4 and 28S ribosomal RNA (rRNA) probes or protein by Western blotting of conditioned media. B, Reverse transcriptase– polymerase chain reaction (RT-PCR) analysis of RNA from control and IL-1–treated human synovial fibroblasts. Results are representative of studies of fibroblasts from 4 different patients. Induction of ADAMTS-4 by interleukin-1: comment on the article by Pratta et al To the Editor: We read with great interest the article by Pratta et al (1), who attributed increased interleukin-1 (IL-1)–stimulated aggrecanase 1 (ADAMTS-4) activity to activation of the constitutively produced enzyme by an activator, and not to its increased production in bovine nasal cartilage, chondrocytes, and capsular fibroblasts. Results from several laboratories support the notion of increased enzyme activity in response to IL-1 stimulation. However, the results of various studies do not support the conclusion about lack of increased ADAMTS-4 production. We therefore conducted experiments to further investigate the inducibility of ADAMTS-4 expression by IL-1. Confluent bovine articular chondrocytes and normal human synovial fibroblasts were maintained in serum-deficient medium for 24 hours and stimulated with recombinant human IL-␤ for 24 hours. ADAMTS-4 RNA was analyzed by reverse transcriptase–polymerase chain reaction (RT-PCR) with specific primers as reported previously (2), or by Northern hybridization with an antisense probe after cloning and sequencing of the RT-PCR product. In contrast to the findings reported by Pratta et al (1), we consistently observed IL-1–inducible expression of the 4.2-kb ADAMTS-4 RNA by Northern blotting (Figure 1A) and RT-PCR analysis (results not shown) in bovine chondrocytes with similar levels of 28S ribosomal or GAPDH RNA. By Western blotting of the conditioned media with an antibody against ADAMTS-4 (Triplepoint Biologics, Forest Grove, OR), we also detected IL-1–inducible 98 kd (zymogen form) and 64-kd (processed form) ADAMTS-4 protein bands (Figure 1A). These results are consistent with those reported by Tortorella et al (3), Little et al (2), and Sztrolovics et al (4). ADAMTS-4 RNA (692-bp complementary DNA [cDNA]) was also shown to be induced by IL-1␤ in RT-PCR analysis of several human synovial fibroblast cell lines from different patients, while control GAPDH RNA levels (226-bp cDNA) remained constant (Figure 1B). Yamanishi and colleagues (5) reported similar IL-1␤–inducible expression of ADAMTS-4 in human synovial fibroblasts. Findings have varied with regard to inducibility in human chondrocytes or cartilage. For example, Flannery et al (6) found no ADAMTS-4 RNA induction by IL-1␤, tumor necrosis factor ␣, or retinoic acid, but the latter increased aggrecanase activity, suggesting a posttranscriptional activation mechanism as supported by Pratta et al (1). In a real-time PCR analysis of human T/C28a4 chondrocytes, Koshy and colleagues (7) did not observe any induction of ADAMTS-4 RNA by IL-1␣. Similarly, Bau et al (8) reported only slight induction of ADAMTS-4 RNA by IL-1␤ in normal human chondrocytes or cells grown in alginate beads. Aggrecanase activity was induced by retinoic acid but not by IL-1␤ in human adult cartilage (4). Interestingly, induction depended upon the type of stimulus, species, and age of the cartilage. Thus, most of the available evidence from studies of bovine chondrocytes supports the notion that ADAMTS-4 induction by IL-1 occurs via both increased production and enhanced activity, in contrast to the findings reported by Pratta et al (1). Basal (constitutive) aggrecanase 1 RNA and protein expression is inducible by IL-1 in bovine articular chondrocytes, cartilage explants, and human synovial fibroblasts. Its inducibility in the arthritis-pertinent model of human adult chondrocytes is controversial and requires additional research. Abdelhamid Liacini, MSc Muhammad Zafarullah, PhD CHUM-Notre-Dame Hospital Montreal, Quebec, Canada LETTERS 1. Pratta MA, Scherle PA, Yang G, Liu RQ, Newton RC. Induction of aggrecanase 1 (ADAM-TS4) by interleukin-1 occurs through activation of constitutively produced protein. Arthritis Rheum 2003;48: 119–33. 2. Little CB, Hughes CE, Curtis CL, Jones SA, Caterson B, Flannery CR. Cyclosporin A inhibition of aggrecanase-mediated proteoglycan catabolism in articular cartilage. Arthritis Rheum 2002;46: 124–9. 3. Tortorella MD, Malfait AM, Deccico C, Arner E. The role of ADAM-TS4 (aggrecanase-1) and ADAM-TS5 (aggrecanase-2) in a model of cartilage degradation. Osteoarthritis Cartilage 2001;9: 539–52. 4. Sztrolovics R, White RJ, Roughley PJ, Mort JS. The mechanism of aggrecan release from cartilage differs with tissue origin and the agent used to stimulate catabolism. Biochem J 2002;362:465–72. 5. Yamanishi Y, Boyle DL, Clark M, Maki RA, Tortorella MD, Arner EC, et al. Expression and regulation of aggrecanase in arthritis: the role of TGF-␤. J Immunol 2002;168:1405–12. 6. Flannery CR, Little CB, Hughes CE, Caterson B. Expression of ADAMTS homologues in articular cartilage. Biochem Biophys Res Commun 1999;260:318–22. 7. Koshy PJ, Lundy CJ, Rowan AD, Porter S, Edwards DR, Hogan A, et al. The modulation of matrix metalloproteinase and ADAM gene expression in human chondrocytes by interleukin-1 and oncostatin M: a time-course study using real-time quantitative reverse transcription–polymerase chain reaction. Arthritis Rheum 2002;46: 961–7. 8. Bau B, Gebhard PM, Haag J, Knorr T, Bartnik E, Aigner T. Relative messenger RNA expression profiling of collagenases and aggrecanases in human articular chondrocytes in vivo and in vitro. Arthritis Rheum 2002;46:2648–57. DOI 10.1002/art.20457 Reply To the Editor: I appreciate the opportunity to discuss the points raised by Liacini and Zafarullah. In our article, we reported that induction of aggrecanase activity involved activation of constituitively expressed protein. We based this conclusion on the observations that 1) unstimulated chondrocytes, isolated articular and nasal cartilage, and capsular fibroblasts contain relatively high levels of cell-associated ADAMTS-4 based on immunocytochemistry studies; and 2) there was no detectable difference in the total ADAMTS-4–immunoreactive proteins present in unstimulated and IL-1–stimulated articular chondrocytes (based on immunofluorescence studies) and cell lysates (based on Western blot analysis). Consistent with conversion from zymogen to active ADAMTS-4, there was an apparent “shift” in the molecular weight of immunoreactive ADAMTS-4 proteins upon IL-1 stimulation, and activity was confirmed in a novel cell-based assay using an immobilized peptide substrate. Specificity of the anti–ADAMTS-4 (antiVMAH) used in the immunofluorescence studies and the Western blot analysis was confirmed by the demonstration that both staining and immunoblotting could be blocked by preincubation of the antibody with immunizing peptide, but not with an irrelevant peptide. Antibody specificity was also confirmed by the demonstration of immunofluorescence of transfected 2039 Drosophila S2 cells, but not of untransfected cells or transfected cells stained in the presence of immunizing peptide. Although we didn’t specifically measure effects on ADAMTS-4 expression in these studies, we are aware that IL-1 can cause an increase in ADAMTS-4 expression, and we described the findings of others in demonstrating this effect. Based on studies reported by others, as well as studies that we performed following publication of our report, we would agree that IL-1 induces a marginal increase in ADAMTS-4 expression, and, as suggested by others including Liacini and Zafarullah, we agree that the increase depends on species and age of the chondrocytes. However, none of these data address the presence of cell-associated ADAMTS-4 in chondrocytes and capsular fibroblasts, and there is no simple way to distinguish the origin of the active enzyme present in our cell-based assay or in IL-1–induced cartilage explants. While it is reasonable to assume that IL-1 induces ADAMTS-4 expression and this leads to a corresponding increase in protein and activity as suggested by Liacini and Zafarullah, the results of our Western blot and immunofluorescence studies do not support this conclusion. Their hypothesis also ignores the fate of the cell-associated ADAMTS-4, and it is reasonable to assume that this population of enzyme could contribute to the activity measured in the cell-based assay and to aggrecan degradation in IL-1–stimulated cartilage explants through induction of an as-yet-unidentified activator that is produced in response to IL-1 stimulation. It is intriguing that Liacini and Zafarullah were able to generate presumably “active” ADAMTS-4 from IL-1– stimulated bovine chondrocytes. We, as well as many others, have never observed a soluble form of the enzyme released from monolayer chondrocytes. Liacini and Zafarullah are incorrect in concluding that their observation with isolated chondrocytes is supported by others (Tortorella et al, Little et al, Sztrolovics et al) since those authors reported induction of ADAMTS-4 in response to IL-1–stimulated cartilage explants, and not by monolayer chondrocytes (1–3). We noted in our report that conditioned medium from IL-1–stimulated bovine chondrocytes was assayed for aggrecanase activity using both the peptide substrate described as well as isolated bovine aggrecan monomers, and we were unable to demonstrate aggrecanase activity in the conditioned medium. Consistent with these results, we were unable to detect the presence of ADAMTS-4 in the conditioned medium by Western blot analysis. For these reasons, we designed the cell-based assay for aggrecanase and were able to detect cell-associated aggrecanase activity, specifically ADAMTS-4 based on the use of selective inhibitors, from monolayer chondrocytes and capsular fibroblasts. Liacini and Zafarullah conclude that the immunoreactive proteins observed in their Western blot analyses represent active ADAMTS-4, based presumably on the size of the proteins. I have several questions regarding this conclusion. Have they demonstrated enzymatic activity of the conditioned medium that contains the proposed active ADAMTS-4? Does this medium digest aggrecan monomers and give rise to the characteristic cleavage products attributed to ADAMTS-4? Have they demonstrated specificity of the antibody used in these studies? Without these appropriate controls, it is difficult to draw any conclusions regarding the identity of the proteins detected by Western blot analysis. 2040 LETTERS These studies continue to raise questions regarding the identity of the activator involved in activation of constituently expressed ADAMTS-4. It has been reported that ADAMTS-4, as well as other ADAM family members, contain a furin activation site, and it is possible that a furin-like protease may be contributing to activation of ADAMTS-4 in these studies. It has also been suggested that a glycosylphosphatidyl inositol– anchored protease, possibly membrane type 4 matrix metalloproteinase, may contribute to ADAMTS-4 activation (4). Identification of the activators will require further investigation and may provide an alternative target for therapeutic intervention in rheumatic diseases. Michael A. Pratta, MS GlaxoSmithKline King of Prussia, PA 1. Tortorella MD, Malfait AM, Deccico C, Arner E. The role of ADAM-TS4 (aggrecanase-1) and ADAM-TS5 (aggrecanase-2) in a model of cartilage degradation. Osteoarthritis Cartilage 2001;9: 539–52. 2. Little CB, Hughes CE, Curtis CL, Jones SA, Caterson B, Flannery CR. Cyclosporin A inhibition of aggrecanase-mediated proteoglycan catabolism in articular cartilage. Arthritis Rheum 2002;46: 124–9. 3. Sztrolovics R, White RJ, Roughley PJ, Mort JS. The mechanism of aggrecan release from cartilage differs with tissue origin and the agent used to stimulate catabolism. Biochem J 2002;362:465–72. 4. Gao G, Plaas A, Thompson VP, Jin S, Zuo F, Sandy JD. ADAMTS4 (aggrecanase-1) activation on the cell surface involves C-terminal cleavage by glycosylphosphatidyl inositol-anchored membrane type 4-matrix metalloproteinase and binding of the activated proteinase to chondroitin sulfate and heparan sulfate on syndecan-1. J Biol Chem 2004;279:10042–51. DOI 10.1002/art.579 Erratum In the article by Eekhoff et al published in the May 2004 issue of Arthritis & Rheumatism (pp 1650–1654), there was an error in the second-to-last sentence of the fourth paragraph of the Results section. The sentence should have read, “The third mutation was a T⬎C conversion at position ⫹1251 [not ‘⫹1250’], resulting in a methionine–threonine substitution at codon 404 (M404T).” We regret the error.