Anti-tumour necrosis factor (TNF)-a therapy
(etanercept) down-regulates serum matrix
metalloproteinase (MMP)-3 and MMP-1 in
A. I. Catrina, J. Lampa, S. Ernestam1, E. af Klint, J. Bratt1,
L. Klareskog and A.-K. Ulfgren
Department of Rheumatology, Karolinska Hospital and1Department of
Rheumatology, Huddinge Hospital, Stockholm, Sweden
Objectives. Matrix metalloproteinases (MMPs) are cytokine-modulated enzymes that play an
important role in the pathogenesis of rheumatoid arthritis (RA) by inducing bone resorption and
cartilage destruction. This study evaluated the modulation of serum and synovial MMPs and
their inhibitor, tissue inhibitor of matrix metalloproteinases (TIMP)-1, by therapy with soluble
tumour necrosis factor (TNF) a receptor (etanercept).
Methods. Serum samples were collected from 60 RA patients at baseline and after 8 or
12 weeks of treatment. Paired synovial biopsies were obtained from 11 patients at two time
points, before and after 8 weeks of treatment. We measured serum levels of MMP-1, MMP-3 and
TIMP-1 by ELISA. Immunohistological analysis of synovial tissue was performed using
monoclonal antibodies specific for MMP-1, MMP-3 and TIMP-1.
Results. Etanercept therapy significantly down-regulated serum levels of MMP-3 and MMP-1
in parallel with the reduction in inflammatory parameters (C-reactive protein concentration and
erythrocyte sedimentation rate) in RA patients. Baseline pretreatment serum levels of MMP-3
correlated with changes in clinical disease activity during therapy. No consistent changes in
serum level of TIMP-1 were observed, while ratios of MMP-1 and MMP-3 to TIMP-1 were
down-regulated following etanercept treatment. Immunohistochemical analyses revealed great
interindividual variability, with generally a high level of expression of MMP and low expression
of TIMP. No significant change in the pattern or number of positive cells occurred during
Conclusions. In RA patients, etanercept therapy down-regulates serum levels of MMP-3 and
MMP-1 and the ratio between MMPs and TIMP-1. This may be an important mechanism for
the prevention of future development of joint damage.
KEY WORDS: Rheumatoid arthritis, Soluble tumour necrosis factor a receptor, Etanercept,
Metalloproteinases, Tissue inhibitors of metalloproteinases.
Rheumatoid arthritis (RA) is an autoimmune inflam-
matory disease characterized by cartilage destruction
and extracellular matrix degradation. Cytokines, such
as tumour necrosis factor (TNF)-a and interleukin
(IL)-1, play a major role in the pathogenesis of the
disease, being able to induce bone resorption and
cartilage destruction (for review see w1x). These two
cytokines act synergistically to release matrix metallo-
proteinases (MMPs), a class of zinc-dependent pep-
tidases participating in extracellular matrix degradation
and remodelling. MMPs are synthesized and secreted
as latent pro-enzymes and their activation is due to
proteolysis of a pro-peptide domain at the N-terminus
of the molecule.
At present there are at least 19 known human MMPs,
which can be divided into four groups: the collagenases,
the stromelysins, the gelatinases and the membrane-
type MMPs (for review see w2x). Stromelysin 1 (MMP-3)
and collagenase (MMP-1) are thought to be probably
involved in the pathogenesis of RA w3x. MMP-3 levels
are increased both in synovial fluid w4x and serum w5x
in RA patients, with a highly significant correlation
between matched samples w6x, suggesting that serum
MMP-3 is derived mainly from that synthesized in the
synovium. MMP-1 (fibroblast collagenase) is detected in
Submitted 1 May 2001; revised version accepted 2 November 2001.
? 2002 British Society for Rheumatology
RA synovial fluid w7x and serum samples w8x at higher
levels than in healthy controls. Tissue inhibitor of matrix
metalloproteinase (TIMP)-1 is a specific inhibitor of
MMPs. It is present in RA synovium to a lesser extent
compared with MMPs, suggesting imbalance between
the two components w9x.
Treatment with TNF blocking agents, alone or in
combination with methotrexate, has high clinical effi-
cacy and delays joint destruction in RA (for review
see w1x). Two TNF blocking agents are available for
clinical use: infliximab (a chimeric monoclonal anti-
body) and etanercept (a recombinant TNF receptor–Fc
fusion protein). Several mechanisms have been pro-
posed to explain the clinical efficacy of infliximab, such
as decreased cell recruitment at the site of inflam-
mation w10x, down-regulation of synovial cytokine
expression w11x and reduction of MMP-1 and MMP-3
synthesis w12x. However, it is not known if etanercept
therapy modulates the serum levels of MMPs and
TIMP, and there are no reports on parallel studies of
MMPs and TIMP in serum and synovial tissue during
TNF blockade either for etanercept or for infliximab.
Materials and methods
Patients and samples
Sixty patients (50 females and 10 males, median age
53 yr, range 30–80 yr) meeting the American College
of Rheumatology (ACR) criteria for RA w13x were
recruited for this study. All patients had a history of
failed therapy with at least one disease-modifying
anti-rheumatic drug (DMARD) at trial entry. No
patient received more than 10 mg prednisolone daily.
All patients received a subcutaneous injection of
25 mg etanercept (Wyeth-Ayerst, Philadelphia, PA,
USA) twice a week. Twenty-five patients were treated
with etanercept alone. Thirty-two patients received
etanercept in combination with methotrexate and cyclo-
sporin, one patient etanercept in combination with
sulphasalazine, and one patient etanercept in combina-
Uppsala, Sweden), an anti-rheumatic agent w14x used on
a licence basis in Sweden.
Serum samples were collected from all 60 patients
before treatment and after 8 weeks of treatment
(23 patients, group 1) or after 12 weeks of treatment
(37 patients, group 2). Serum samples were stored
at 2208C until assayed. Synovial biopsies were obtained
at arthroscopy from 11 patients in group 1 before
treatment and after 8 weeks of treatment.
Clinical and laboratory assessments
All patients were assessed for overall activity of the
disease using the (disease activity score) DAS28 w15x
before treatment and after 12 weeks of treatment.
Changes in DAS28 score during therapy were calculated
as the difference between the estimated value after
12 weeks of treatment and the baseline pretreatment
value. The ACR response w16x to therapy was also
recorded after 12 weeks of treatment for all 60 patients.
Laboratory parameters, namely C-reactive protein
(CRP) concentration and erythrocyte sedimentation
rate (ESR), were recorded before and after 12 weeks
of treatment for all 60 patients.
Serum levels of MMP-1, MMP-3 and TIMP-1 were
(ELISA) (Amersham Pharmacia Biotech, Uppsala,
Sweden) according to the manufacturer’s directions.
Serum samples were diluted (1:1 for MMP-1, between
1:1 and 1:8 for MMP-3 and between 1:10 and 1:20 for
TIMP-1) and individual samples were assayed in duplic-
ate. The MMP-1 kit recognized total human MMP-1,
free MMP-1 and MMP-1 complexed with TIMP-1, but
not MMP-1 complexed with a2-macroglobulin. The
range of the assay was 6.25–100 nguml with sensitivity of
1.7 nguml. The MMP-3 assay recognized proMMP-3,
active MMP-3 and MMP-3uTIMP complexes but not
MMP-3 bound by a2-macroglobulin, in the range
3.75–120 nguml with sensitivity of 2.35 nguml. Total
TIMP-1 (free TIMP-1 and TMP-1 complexed with
MMPs) was recognized by ELISA in the range
3.13–50 nguml with sensitivity of 1.25 nguml. Ratios
of MMP-3 to TIMP-1 and MMP-1 to TIMP-1 were
Tissue preparation and immunohistochemical analysis
Eleven patients from group 1 were subjected to knee
arthroscopy before treatment and after 8 weeks of
treatment. Biopsy specimens adjacent to the cartilage–
pannus junction were snap-frozen in liquid nitrogen
and then kept at 2708C until sectioned. Serial cryostat
sections (7 mm) were fixed for 20 min with 2% (vuv)
formaldehyde (Sigma Chemicals, St Louis, MO, USA)
and stored at 2708C. The staining procedure has been
published previously w17x. We used the following mouse
monoclonal antibodies (all purchased from Oncogene,
Cambridge, MA, USA): anti-MMP-1 IgG2a (41-1E5),
anti-MMP-3 IgG1 (55-2A4) and anti-TIMP-1 IgG1
(7-6C1). Negative controls with matched IgG isotype
and recombinant protein-blocked
included for each marker w17x.
Stained synovial biopsy sections were evaluated
using computerized image analysis as described pre-
viously w18x. The areas of MMP and TIMP staining were
expressed as percentages of the total area of counter-
stained tissue. Analysis of an entire tissue section
typically involved 30–100 microscopic fields at a
magnification of 3250.
Differences between groups were analysed using the
Mann–Whitney U-test and analysis for matched pairs
was performed using Wilcoxon’s signed rank test.
Correlations between variables were assessed using
the Spearman rank correlation test. P<0.05 was
considered statistically significant.
Etanercept down-regulates MMP-3 and MMP-1 in RA
After 12 weeks of treatment, 47 of 60 patients (78.33%)
fulfilled ACR20 criteria, while 13 of 60 patients (21.66%)
showed no response to therapy. Clinical and laboratory
parameters improved significantly after 12 weeks of
treatment in all 60 patients, with reduction of the
DAS28 score wfrom 6.01 to 4.02 (median values),
P<0.0001x, CRP level wfrom 34 mguml to 15 mguml
(median values), P<0.0001x and ESR wfrom 42 to
27 mmuh (median values), P<0.0001x.
Serum levels of MMPs and TIMPs
Table 1 summarizes the results of ELISA analyses.
Following treatment, the median serum level of MMP-3
decreased significantly in group 1 of patients from 68.1
to 46.5 nguml (P<0.001) and in group 2 of patients
from 115.3 to 48.8 nguml (P<0.0001). The median
serum level of MMP-1 decreased significantly in group 1
patients from 9.3 to 8.7 nguml (P<0.05) and in group 2
patients from 9.8 to 7.9 nguml (P=0.001). The median
serum level of TIMP-1 was down-regulated following
8 weeks of treatment for patients in group 1 (from 433.2
to 371.6 nguml, P<0.05) but remained unchanged
compared with the baseline value after 12 weeks for
patients in group 2 (Table 1 and Fig. 1). In group 1, the
MMP-1:TIMP-1 and MMP-3:TIMP-1 ratios were not
significantly changed after 8 weeks of treatment. In
group 2, both the median MMP-1:TIMP-1 ratio and the
median MMP-3:TIMP-1 ratio decreased significantly
wfrom 0.025 to 0.020 (P<0.05) for MMP-1:TIMP-1
(Table 1 and Fig. 2).
We did not observe any difference between cortisone-
treated and non-treated patients or between patients
treated with DMARDs and those not treated with
TABLE 1. Serum levels of MMP-3, MMP-1 and TIMP-1 and their ratios before treatment (week 0) and after 8 weeks of treatment (group 1) or after
12 weeks of treatment (group 2)
Group 1 Group 2
Week 0 Week 8 Week 0 Week 12
Values are median (range).
FIG. 1. Changes in serum levels of MMP-3 (A), MMP-1 (B)
and TIMP-1 (C) during etanercept treatment in group 1 (white
boxes) and group 2 (grey boxes). Horizontal lines represent
medians and whiskers the non-outlier values. *P<0.05;
**P<0.01; ***P<0.001 (non-parametric Mann–Whitney
A. I. Catrina et al.
Correlations between serum levels of MMPs and
TIMP-1 and disease parameters
In all 60 patients, the serum level of MMP-3 before
treatment correlated with the baseline CRP level
(P<0.001, r=0.42) and baseline ESR (P<0.01,
r=0.38) (data not shown). Moreover, the baseline
serum level of MMP-3 correlated with the change in
FIG. 2. ChangesinMMP-1:TIMP-1(A)andMMP-3:TIMP-1
(B) ratios during etanercept treatment in group 1 (white boxes)
and group 2 (grey boxes). Horizontal lines represent medians
and whiskers the non-outlier values. *P<0.05; ***P<0.001
(non-parametric Mann–Whitney U-test).
TABLE 2. Synovial expression of MMP-1, MMP-3 and TIMP-1 before
treatment and after 8 weeks of etanercept treatment in 11 patients
Positive cells (%)
Week 0Week 8
Values are median percentage of positive cells (percentage of
positively stained area as a percentage of total counterstained area)
evaluated by computer image analysis, with ranges in parentheses.
FIG. 3. Photographs illustrating brown immunoperoxidase staining for MMPs and TIMP-1 in serial sections from cryopreserved
synovial biopsies counterstained with haematoxylin. (A) MMP-1. (B) MMP-3. (C) TIMP-1. Original magnification 3310. All
sections are from the same patient before treatment.
Etanercept down-regulates MMP-3 and MMP-1 in RA
DAS score after 12 weeks of treatment (P<0.05,
r=20.28) (data not shown).
Pretreatment serum level of TIMP-1 correlated with
baseline CRP level (P<0.001, r=0.43) and ESR
(P<0.01, r=0.34) (data not shown). Pretreatment
serum level of MMP-1 did not correlate with any
Table 2 summarizes the results of immunohistochemical
analysis of the biopsies obtained from 11 patients in
group 1 at baseline and after 8 weeks of treatment. We
observed great inter-individual variability even though
most patients had a high level of MMP expression and a
low level of TIMP expression (Fig. 3). Matrix metallo-
proteinases were present in all biopsies studied, both
before and after treatment, while TIMP-1 was identified
in 20 out of 22 biopsies investigated.
MMP-3 was present mainly in the lining and endo-
thelial cells. MMP-1 was expressed mainly in the sub-
lining layer and around vessels and to a lesser extent in
the lining layer. TIMP-1 showed a generally low level
of expression, with positive signal in isolated areas.
No significant changes in the pattern or number of
positive cells occurred after 8 weeks of treatment.
MMPs have a major role in RA pathogenesis as
mediators of cartilage and bone destruction, and this
is the first report describing modulation of the MMP–
TIMP system by soluble TNF-a receptors (etanercept)
in RA patients.
We focused our study on MMP-3 and MMP-1,
measuring their serum levels with ELISA kits that
recognized both the active and inactive forms of MMPs.
Serum levels of MMP-3 quantified by this method
have been shown previously to correlate with radio-
logical damage w19–20x in RA patients. We found that
etanercept therapy significantly decreased the serum
level of MMP-3 as well as the ratio of MMP-3 to
TIMP-1. These changes could explain the ability of anti-
TNF therapy to prevent further development of joint
damage. Serum levels of MMP-3 correlated highly with
inflammatory markers (CRP and ESR), as shown pre-
viously w8, 21x. Moreover, our data indicate that baseline
MMP-3 serum levels may predict the changes in DAS
score during therapy.
We also observed a decrease in the serum level of
MMP-1 during etanercept therapy, with reduction of the
MMP-1:TIMP-1 ratio. It was shown previously that
the serum level of MMP-1 correlated with the number
of new erosions that developed in RA patients over
18 months of follow-up w22x. Thus, the decrease in the
serum level of MMP-1 could partly explain the delay
in radiological progression in patients with RA during
etanercept treatment. The serum level of MMP-1 did not
correlate with inflammatory parameters, suggesting that
the effect of etanercept on the serum level of MMP-1 is
independent of the down-regulation of inflammation.
The serum level of TIMP-1 before treatment correl-
ated with both CRP and ESR. Interestingly, etanercept
therapy did not influence the serum level of TIMP-1 con-
sistently despite a general decrease in all inflammatory
markers investigated. This finding suggests that TIMP-1
is influenced mainly by a pro- anduor anti-inflammatory
mechanism other than one involving TNF.
Several reports have suggested that there might be
an imbalance between MMP and TIMP expression in
the RA synovium w3x, allowing MMP activation and
the subsequent destruction of joint components. A
recent study reported that a high level of synovial
expression of MMP-1 mRNA in early RA distinguished
patients with more rapidly progressive erosive disease
w23x. We did not observe any changes in the synovial
expression of MMPs and TIMP-1 during etanercept
therapy, however, even though biopsies were taken
from areas adjacent to the cartilage pannus junction.
Possible explanations for this finding are the low
number of patients subjected to arthroscopy anduor
lack of specificity between the recognition of active
and inactive forms of MMPs.
In conclusion, we have demonstrated that therapy
with soluble TNF-a receptors significantly decreases the
serum levels of MMP-3 and MMP-1, with reductions
in the ratios of MMPs to TIMP-1. The pretreatment
serum level of MMP-3 may be considered a potential
predictor of changes in clinical disease activity during
treatment with etanercept.
We thank Associate Professor R. A. Harris for linguistic
advice. Marianne Engstrom provided excellent technical
assistance. The study was supported by grants from
the Swedish Medical Research Council, the Swedish
Rheumatism Association, the insurance company AMF
and from Freemasan Barnhuset Stockholm.
rheumatoid arthritis. Annu Rev Med 2000;51:207–29.
2. Curran S, Murray GI. Matrix metalloproteinases in
tumour invasion and metastasis. J Pathol 1999;189:300–8.
3. Firestein GS. Etiology and pathogenesis of rheumatoid
arthritis. In: Kelley WN, Harris ED Jr, Ruddy S,
Sledge CB, eds. Textbook of Rheumatology, Vol. 1, 5th
edn. Philadelphia: W.B. Saunders, 1997:851–97.
4. Ishiguro N, Ito T, Obata K, Fujimoto N, Iwata H. Deter-
mination of stromelysin-1, 72 and 92 kDatype IVcollagen-
ase, tissue inhibitor of metalloproteinase-1 (TIMP-1), and
TIMP-2 in synovial fluid and serum from patients with
rheumatoid arthritis. J Rheumatol 1996;23:1599–604.
5. Zucker S, Lysik RM, Zarrabi MH et al. Elevated plasma
stromelysin levels in arthritis wsee commentsx. J Rheumatol
6. Yoshihara Y, Obata K, Fujimoto N, Yamashita K,
stromelysin-1 and tissue inhibitor of metalloproteinases-1
in sera from patients with rheumatoid arthritis. Arthritis
RN,Taylor PC. Anti-cytokinetherapy for
A. I. Catrina et al.
7. Clark IM, Powell LK, Ramsey S, Hazleman BL,
Cawston TE. The measurement of collagenase, tissue
inhibitor of metalloproteinases (TIMP), and collagenase–
TIMP complex in synovial fluids from patients with osteo-
arthritis and rheumatoid arthritis. Arthritis Rheum 1993;
8. Keyszer G, Lambiri I, Nagel R et al. Circulating levels
of matrix metalloproteinases MMP-3 and MMP-1, tissue
inhibitor of metalloproteinases 1 (TIMP-1), and MMP-1u
TIMP-1 complex in rheumatic disease. Correlation with
clinical activity of rheumatoid arthritis versus other
surrogate markers wsee commentsx. J Rheumatol 1999;
9. Firestein GS, Paine MM, Littman BH. Gene expression
(collagenase, tissue inhibitor of metalloproteinases, com-
plement and HLA-DR) in rheumatoid arthritis and osteo-
arthritis synovium. Quantitative analysis and effect of
10. Tak PP, Taylor PC, Breedveld FC et al. Decrease in
cellularity and expression of adhesion molecules by anti-
tumor necrosis factor alpha monoclonal antibody treat-
ment in patients with rheumatoid arthritis wsee commentsx.
Arthritis Rheum 1996;39:1077–81.
11. Ulfgren AK, Andersson U, Engstrom M, Klareskog L,
Maini RN, Taylor PC. Systemic anti-tumor necrosis factor
alpha therapy in rheumatoid arthritis down-regulates
synovial tumor necrosis factor alpha synthesis. Arthritis
12. Brennan FM, Browne KA, Green PA, Jaspar JM,
Maini RN, Feldmann M. Reduction of serum matrix
metalloproteinase 1 and matrix metalloproteinase 3 in
rheumatoid arthritis patients following anti-tumour nec-
rosis factor-alpha (cA2) therapy. Br J Rheumatol 1997;
13. Arnett FC, Edworthy SM, Bloch DA et al. The American
Rheumatism Association 1987 revised criteria for the
classification of rheumatoid arthritis. Arthritis Rheum
14. Larsen A, Petersson I, Svensson B. Podophyllum deriv-
atives (CPH 82) compared with placebo in the treatment
of rheumatoid arthritis. Br J Rheumatol 1989;28:124–7.
15. Prevoo ML, van’t Hof MA, Kuper HH, van Leeuwen MA,
van de Putte LB, van Riel PL. Modified disease activity
scores that include twenty-eight-joint counts. Develop-
ment and validation in a prospective longitudinal study
of patients with rheumatoid arthritis wsee commentsx.
Arthritis Rheum 1995;38:44–8.
16. Felson DT, Anderson JJ, Boers M et al. The American
College of Rheumatology preliminary core set of disease
activity measures for rheumatoid arthritis clinical trials.
The Committee on Outcome Measures in Rheumatoid
Arthritis Clinical Trials. Arthritis Rheum 1993;36:729–40.
17. Ulfgren AK, Lindblad S, Klareskog L, Andersson J,
Andersson U. Detection of cytokine producing cells in
the synovial membrane from patients with rheumatoid
arthritis. Ann Rheum Dis 1995;54:654–61.
18. Ulfgren AK, Grondal L, Lindblad S et al. Interindividual
and intra-articular variation of proinflammatory cyto-
kines in patients with rheumatoid arthritis: potential
implications for treatment. Ann Rheum Dis 2000;
19. Posthumus MD, Limburg PC, Westra J et al. Serum levels
of matrix metalloproteinase-3 in relation to the develop-
ment of radiological damage in patients with early
rheumatoid arthritis. Rheumatology 1999;38:1081–7.
20. Yamanaka H, Matsuda Y, Tanaka M et al. Serum matrix
metalloproteinase 3 as a predictor of the degree of joint
destruction during the six months after measurement, in
patients with early rheumatoid arthritis. Arthritis Rheum
Uchiyama M. Serum matrix metalloproteinase-3 and
fibrin degradation product levels correlate with clinical
diseaseactivity in rheumatoid
22. Cunnane GF, Fitzgerald O, Summers CA, Cawston T,
Bresnihan B. Early joint erosions and serum levels of
matrix metalloproteinase (MMP)-1. MMP-3 and tissue
inhibitor of metalloproteinase-1 (TIMP) in rheumatoid
arthritis. Arthritis Rheum 2000;43:S67.
23. Cunnane G, FitzGerald O, Hummel KM et al. Synovial
tissue protease gene expression and joint erosions in early
rheumatoid arthritis. Arthritis Rheum 2001;44:1744–53.
Etanercept down-regulates MMP-3 and MMP-1 in RA