Rheumatology 2000;39:274–282
What determines arthritogenicity of bacterial
cell wall? A study on Eubacterium cell wallinduced arthritis
X. Zhang, M. Rimpiläinen, E. Šimelyte and P. Toivanen
Turku Immunology Centre, Department of Medical Microbiology,
Turku University, Turku, Finland
Abstract
Objective. To study what determines the arthritogenicity of the bacterial cell wall (CW )
using Eubacterium CW-induced arthritis in the rat.
Methods. Eubacterium aerofaciens, previously reported as arthritogenic, and E. limosum and
E. alactolyticum, known as non-arthritogenic, were used. Gas chromatography–mass
spectrometry (GC–MS ) was applied to analyse the chemical composition of the bacterial cell
wall. Cellular immune response was measured by concanavalin A (Con A) stimulation and
FACScan analysis. Also, serum antibodies against the injected cell wall were determined.
Results. Unexpectedly, from the two strains of E. aerofaciens used only one proved to be
arthritogenic (with a CW inducing chronic arthritis after a single intraperitoneal injection),
even though these two strains were 100% identical by 16S rDNA analysis. CW of the other
E. aerofaciens strain induced only transient acute arthritis; CW of E. limosum and
E. alactolyticum induced weak signs of acute arthritis. Based on the GC–MS analysis and on
the results published previously, putative structures of peptidoglycan (PG) in the four CW
preparations are presented. It is apparent that the presence of lysine in position 3 of the PG
stem peptide contributes to arthritogenicity but is alone not decisive. Both strains of
E. aerofaciens were immunosuppressive, when tested by Con A response at 2 weeks after
CW injection. Such an immunosuppression was not observed after injection of CW from
E. limosum or E. alactolyticum. FACScan analysis for six T cell markers and studies on serum
antibody responses did not reveal any differences in the effect of the four bacterial strains
used.
Conclusions. The results obtained suggest that the chemical structure of PG present in the
bacterial CW is decisive in determining arthritogenicity/non-arthritogenicity. Therefore, from
two bacterial strains belonging to normal human intestinal flora and 100% identical by 16S
rDNA analysis, one proved to be arthritogenic and the other non-arthritogenic.
K : Intestinal flora, Peptidoglycan, Gas chromatography, Mass spectrometry, Lysine.
From the animal models of rheumatoid arthritis,
those based on the use of a single i.p. injection
of Gram-positive cell walls (CW ) into rats, are
among the most popular. The bacteria tried in these
experiments include Streptococcus, Lactobacillus,
Bifidobacterium,
Clostridium,
Peptostreptococcus,
Coprococcus, Propionibacterium, and Eubacterium [1–5].
In Eubacterium CW-induced arthritis, histopathological
findings of the joint inflammation closely resemble those
observed in human rheumatoid arthritis, including synovial infiltration by T lymphocytes, synovial lining cell
hyperplasia, pannus formation, tendonitis, and cartilage
and bone erosions [4, 6 ]. Likewise, the presence of T
helper cells and macrophages in the synovial tissue has
been reported [7], and even autoreactive T cell lines
which can behave as arthritogenic have been described
[8].
A major constituent of the Gram-positive CW is the
peptidoglycan–polysaccharide (PG–PS ) complex, which
is known to initiate and sustain a chronic destructive
inflammation. This capacity is dependent upon resistance of PG–PS to biodegradation and on the consequent
persistence in the tissues [4, 9–12]. The PG moiety of
the PG–PS complex possesses multiple immunological
activities [13, 14]. The inflammatory properties of
PG are decreased after degradation by lysozyme,
N-acetylmuramyl--alanine amidase [15, 16 ], or mutanolysin [17], implying that PG has an important role in
the pathogenesis of the chronic inflammation. The PS
moiety, covalently bound to the PG, seems also to
possess relevant biological activity; streptococcal PG–PS
Submitted 2 June 1999; revised version accepted 20 September 1999.
Correspondence to: X. Zhang, Department of Medical
Microbiology, Turku University, Kiinamyllynkatu 13, FIN-20520
Turku, Finland.
274
© 2000 British Society for Rheumatology
Eubacterium cell wall-induced arthritis
could not induce chronic arthritis if the complex was
separated into PG and PS [18, 19], indicating that the
inflammatory PG might be easily degraded in the
absence of PS.
The composition of PG varies from one bacterial
species to another and even between strains within a
single species. PG consists of several layers (up to 70)
of adjacent N-acetylglucosamine and N-acetylmuramic
acid molecules; the layers are bound by peptides to each
other in a regular fashion. Peptides comprising four to
five amino acids (called stem peptides) are attached to
N-acetylmuramic acid to form the glycan backbone of
PG. Backbones are covalently bound to each other by
interpeptide bridges which may or may not contain one
or more additional amino acid [20]. The structure of
PG is schematically depicted in Fig. 1. The greatest
variation of amino acids in the stem peptide occurs in
position 3. Variation also occurs in the mode of crosslinkage and in the interpeptide bridge. According to the
mode of cross-linkage, PG is divided into group A
(cross-linkage between positions 3 and 4 of the stem
peptide) and the less frequent group B (cross-linkage
between positions 2 and 4). Further division into subgroups is made according to variation of the interpeptide
bridge. In group A, the amino acid linked to muramic
acid is always -alanine, followed by -glutamic acid in
position 2, to form the minimal immunoadjuvant structure muramyl dipeptide (MDP). In group B, the amino
acid linked to muramic acid is usually glycine or serine,
followed by -glutamic acid in position 2.
Eubacterium CW-induced arthritis is of special interest, since E. aerofaciens of human intestine was among
the top 10 bacterial species (out of 54) which distinguished patients with early rheumatoid arthritis from con-
275
trols [21]. In addition, it is the only one of those 10 in
which the CW has been used for arthritis induction.
However, CWs of all Eubacterium species are not arthritogenic. For instance, E. limosum CW is not arthritogenic [22]. Even though resistance to biodegradation
and persistence in the tissues seem to be important for
arthritogenicity, it has remained unknown what characteristics of the CW itself determine these properties, as
well as the capacity for arthritis induction. Burroughs
et al. [23], working with Haemophilus influenzae, demonstrated a structure–activity relationship for the inflammatory properties of PG degradation products.
According to their findings, the pro-inflammatory activity is highly dependent on structural variation, both in
the stem peptide and the cross-linkage. Therefore, it is
possible that, within Eubacterium PG, variation of the
stem peptides and cross-linkage leads to different
responses by the host immune system. In the present
work, we aimed to study these questions by using the
CW of four Eubacterium strains. For this purpose, we
chose a strain of E. aerofaciens proven to be arthritogenic (causing chronic arthritis) and a strain of
E. limosum known as non-arthritogenic [22]. Both of
these species belong to the normal intestinal flora in
man. Additionally, we used another strain of E. aerofaciens and a strain of E. alactolyticum, presuming they
were arthritogenic and non-arthritogenic, respectively;
E. alactolyticum is found in human dental plaque.
Materials and methods
Bacteria
Eubacterium aerofaciens ATCC 25986 was obtained
from the Culture Collection, University of Gothenburg,
F. 1. Schematic picture of PG. The figures indicate the position of amino acids in the stem peptide. M, N-acetylmuramic acid;
N, N-acetylglucosamine; Ala, alanine; Glu, glutamic acid; Orn, ornithine; Dap, diaminopimelic acid; Ser, serine; Lys, lysine.
276
X. Zhang et al.
Sweden, with a nomination CCUG 28087. Eubacterium
aerofaciens ATCC 35085, E. limosum ATCC 8486 and
E. alactolyticum ATCC 17927 (now recognized as
Pseudoramibacter alactolyticus [24]) were purchased
from the American Type Culture Collection, Rockville,
MD, USA. All strains were grown overnight under
strictly anaerobic conditions at 37°C in BBL Schaedler
Broth (Becton Dickinson, MD, USA) to the late logarithmic phase. The strains were characterized by 16S
rDNA sequence analysis as described by Jalava et al.
[25], by biochemical reactions [26 ] and by gas chromatographic (GC ) analysis of cellular fatty acids (CFA) as
described by Eerola and Lehtonen [21, 27]. CW of
E. aerofaciens ATCC 25986 and E. limosum ATCC 8486
have been proven to be arthritogenic and nonarthritogenic, respectively [22]. Arthritis induction by
the other two strains has not been previously studied.
CW preparation
The bacterial CW was isolated as described previously
[18]. Briefly, the cells were harvested by centrifugation,
washed twice with phosphate-buffered saline (PBS) at
pH 7.2, heated at 80°C for 30 min to inactivate autolytic
enzymes [5], and disrupted with glass beads
(Ø 0.45–0.50 mm) in a MSK Cell Homogenizer (B.
Braun Melsungen AG, Melsungen, Germany). The CWs
were collected by centrifugation (Sorvall RC5C, Du
Pont, Wilmington, USA) at 10 000 g, +4°C, for 30 min,
treated with deoxyribonuclease I (1 mg/g wet weight),
ribonuclease (10 mg/g wet weight) (both enzymes from
Sigma Chemical Co., MO, USA) and trypsin (20 mg/g
wet weight; Fluka Chemika AG, Buchs SG,
Switzerland), washed twice with PBS and once with
distilled water, and sonicated in an ice bath for 75 min
(Branson Sonifier, Smith Kline Co., Danbury, CT,
USA). The sonicated CW suspension was centrifuged
at 10 000 g, +4°C, for 20 min. The supernatant
was centrifuged by ultracentrifugation (Sorvall
Ultracentrifuge OTD65B, rotor 60 Ti, Du Pont) at
100 000 g, +4°C, for 60 min. The pellet containing the
CW called 100p, for a precipitate obtained by centrifugation at 100 000g was suspended in water, dialysed
(MWCO 3500, Spectrum, CA, USA) against distilled
water at +4°C for 4 days by changing the water several
times and lyophilized for the chemical analysis. For i.p.
injection into rats, the CW preparation was suspended
in PBS and heated at 90°C for 30 min [28]. The sterility
of the preparation was checked by plating on agar plates
at 37°C and room temperature under aerobic and anaerobic conditions; no bacterial growth was detected after
2 days of culture. The endotoxin tests by E-TOXATE
(Sigma) were also found to be negative.
GC–mass spectrometry (GC–MS)
GC–MS was used to analyse the chemical composition
of the CW preparations [29]. The derivatized molecules
were ionized by the electron impact method and analysed
in the selected ion monitoring (SIM ) mode using single
positive ions at a mass-to-charge ratio (m/z). Sugars
were analysed as alditol acetate derivatives with fucose,
allose, and N-methyl-glucamine (Sigma) as internal
standards. The ions monitored were the same as those
described by Gilbart et al. [29], the only exception was
our use of m/z 289 for rhamnose. Amino acids were
analysed as butyl heptafluorobutyl derivatives using
-norleucine, -methionine and -tryptophan (Sigma) as
internal standards. The ions were selected as described
by Gilbart et al. [29]. One microlitre of the derivative
was injected in the splitless mode and analysed by GC
(model HP 5890A; Hewlett-Packard, Wilmington, DE,
USA) equipped with a fused silica capillary column
(SE-54; Nordian Instruments, Helsinki, Finland) and
coupled directly with a TRIO-1 mass spectrometer ( VG
Instruments, Manchester, UK ). For sugars, the column
oven temperature, started at 50°C, was programmed to
270°C at the rate of 10°C/min and held for 1 min.
Finally, the column was heated to 290°C and held for
5 min. For amino acids, the column oven temperature,
started at 85°C, was programmed to 280°C at the rate
of 10°C/min, held for 1 min, and finally heated to 290°C
and held for 5 min.
Animals and induction of arthritis
Inbred female Lewis rats weighing approximately 150 g
were purchased from Harlan Sprague–Dawley, IN,
USA. Arthritis was induced by i.p. injection of CW
(200 mg dry weight/g rat body weight) suspended in
sterile PBS. Control rats were injected with an equal
volume of sterile PBS. To monitor the development of
arthritis, each limb was assigned a score of 0–4, based
on the degree of erythema, oedema, painfulness and
functional disorder of the ankle and metatarsal joints
(wrist and metacarpal joints), by two independent observers, as described previously [30]. Such an evaluation
has been widely used by us; the results are parallel to
those by histological grading [4, 30–32].
Lymphocyte stimulation assay
To determine the cellular reactivity to concanavalin A
(Con A), rat splenocytes were isolated by using
Lympholyte-Rat (Cedarlane Laboratories, Ontario,
Canada) at 2 weeks after the CW injection. The splenocytes were incubated with Con A (1.25 mg/ml ) in roundbottom microtitre wells (Nunclon, Roskilde, Denmark)
containing 1 × 105 cells/0.2 ml culture medium. The cells
were pulsed with 1 mCi [3H ] thymidine 18 h before
harvesting on day 3 with an automatic cell harvester
(Harvest 96, Tomtec, Germany); the incorporated radioactivity was counted in a beta counter ( Wallac, Turku,
Finland). All samples were assayed in quadruplicate.
FACScan analysis
Fluorescein isothiocyanate ( FITC )- or phycoerythrin
(PE )-conjugated monoclonal antibodies OX-52, OX-35,
OX-40, OX-39, R73, V65 and isotype-matched mouse
IgG (PharMingen, San Diego, CA, USA) were used.
For double-colour staining, splenocytes isolated by
Lympholyte-Rat 2 weeks after CW injection were incubated with FITC-conjugated monoclonal antibodies followed by PE-conjugated monoclonal antibodies at
Eubacterium cell wall-induced arthritis
277
+4°C for 30 min in the dark. After washing twice with
PBS supplemented with 2% fetal calf serum containing
0.01% sodium azide, labelled lymphocytes were analysed
using FACScan (Becton Dickinson, CA, USA) flow
cytometer and CELLquest software. To quantify the
expression of the surface antigens, the mean fluorescence
intensity for each staining was measured.
Antibody assays
Sera for antibody assays were collected by cardiac
puncture 2 weeks after the CW injection. Serum IgM,
IgG, and IgA class antibodies specific for CW were
quantified by a modification of a previously published
enzyme-linked immunosorbent assay ( ELISA) [32]. The
water extract of sterile rat food pellets was used as a
control antigen. The food pellets were composed of
wheat, barley, soy, wheat, fish powder, minerals and
vitamins with a minimal contamination of bacterial
structures (Lactamin AB, Stockholm, Sweden). The
pellet was smashed and dissolved in sterile water; after
low-speed centrifugation, the supernatant was collected
and the protein concentration was measured by the
Lowry protein assay. The supernatant was used as the
antigen. Dynatech 96 microtitre plates (Nunc) were
coated with CW (equivalent 5 mg rhamnose CW/well )
or with rat food antigens (5 mg protein/well ) in PBS.
Rat sera diluted 1:200 were incubated with alkaline
phosphatase-conjugated sheep anti-rat IgG 1:500 ( The
Binding Site Limited, Birmingham, UK ), or with unconjugated mouse anti-rat IgA 1:20 000, or with unconjugated mouse anti-rat IgM 1:20 000 (Zymed Laboratories,
CA, USA). For IgA and IgM detection, alkaline phosphatase-conjugated goat anti-mouse IgG + IgM 1:2000
(Caltag Laboratories, CA, USA) was used as a second
antibody. The absorbances were measured at a wavelength of 405 nm using a Titertek Multiscan plus spectrophotometer (Labsystems, Helsinki, Finland ).
Results
Induction of arthritis
As expected, i.p. injection of CW from E. limosum or
E. alactolyticum caused extremely slight signs of acute
arthritis and no chronic arthritis (Fig. 2). However,
from the two strains of E. aerofaciens, only CW of
ATCC 25986 induced chronic arthritis as expected,
whereas i.p. injection of CW from the strain ATCC
35085 resulted only in acute arthritis, which subsided in
about 10 days without any evidence for chronicity. On
this basis and for the sake of simplicity, E. aerofaciens
ATCC 25986 is called the arthritogenic strain and the
strain ATCC 35085 as well as the strains of E. limosum
and E. alactolyticum are designated non-arthritogenic.
Identification of bacteria
Due to the unexpected findings with the other strain
(ATCC 35085) of E. aerofaciens, the identity of all four
strains at the DNA level was studied. For this purpose,
genes coding for 16S ribosomal RNA were sequenced
(320–452 base pairs). The results obtained indicated
100.0% identity between the two strains of E. aerofaciens,
F. 2. Arthritis development in rats injected i.p. with CWs
from four different strains of Eubacterium. Each symbol represents the mean ± standard error of the mean (...) of five to
12 rats.
78.6% identity between E. aerofaciens and E. limosum
or between E. aerofaciens and E. alactolyticum.
Similarity between E. limosum and E. alactolyticum was
92.6%. Despite the complete identity in this DNA analysis, the two strains of E. aerofaciens were slightly
different both in their biochemical and CFA profiles.
The main differences in the biochemical reactions were
in the acid production from glucose, lactose, mannitol,
rhamnose and salicin. In the CFA profiles, the main
differences were in the relative amounts of the fatty acid
methyl esters 12:0, 16:0, 18:1-cis-9, 18:0 and in dimethyl
acetates 11:0 and 14:0. These results indicate that the
two strains of E. aerofaciens do not represent the same
bacterial clone.
Chemical analysis of CW
The results of the chemical analysis of the four CW
preparations are shown in Table 1. All four preparations
X. Zhang et al.
278
T 1. Chemical analysis of the CW preparations by GC–MS
Component
Carbohydrates
Sugars
Rhamnose
Glucose + galactose
Aminosugars
N-acetylmuramic acid
N-acetylglucosamine
N-acetylgalactosamine
Amino acids
Alanine
Glutamic acid/glutamine
Lysine
Aspartic acid/asparagine
Ornithine
Serine
Diaminopimelic acid
Leucine
Isoleucine
Valine
Phenyalanine
Proline
Tyrosine
Cystine
Threonine
Glycine
E. aerofaciens
ATCC 25986
E. aerofaciens
ATCC 35085
E. limosum
ATCC 8486
E. alactolyticum
ATCC 17927
18.5
30.9
17.1
10.4
20.4
14.9
7.2
38.3
1.9
7.6
–
5.6
6.8
–
2.9
2.1
3.9
5.3
4.7
2.0
5.0
6.0
9.7
4.8
<0.1
<0.1
–
0.4
0.4
0.4
0.2
0.3
0.3
–
–
–
2.5
12.5
0.7
5.7
9.4
0.5
–
0.3
0.6
0.3
0.4
0.4
0.3
<0.1
0.6
0.7
3.6
8.0
9.0
2.7
12.1
3.4
–
2.5
2.0
1.6
1.4
1.4
1.0
–
–
–
7.7
17.3
0.4
0.9
0.2
0.9
3.9
0.3
0.6
0.4
0.7
0.5
0.3
0.1
2.0
1.7
Each component is expressed as % of the dry weight of CW. Figures in bold indicate PG amino sugars and amino acids. Norleucine,
methionine, tryptophan were applied as internal standards for amino acid analysis. Cysteine, arginine and histidine were not observed in any of
the four CW preparations.
had a significant amount (7.2–20.4% of dry weight) of
rhamnose which has been claimed to be important for
arthritogenicity [33]. Of interest was the finding that
CW of E. limosum and E. alactolyticum, which were
extremely mild in the acute arthritogenicity, had a small
amount (2.0–3.9%) of N-acetylgalactosamine, whereas
CWs of both strains of E. aerofaciens were completely
devoid of this compound. The total amount of carbohydrates in the four CWs varied from 39.9 to 58.9%
and that of amino acids from 27.7 to 48.7%. Regarding
the quantitative content of PG components N-acetylglucosamine and N-acetylmuramic acid, no such differences
between the four preparations were observed that could
be considered significant for arthritogenicity/nonarthritogenicity.
The amino acids detected ( Table 1) can be divided
into those known to be part of PG and those belonging
to the proteins outside PG. The latter ones are intimately
attached to PG, and their complete separation from the
CW preparations would not have been possible without
breaking the PG structure. Therefore, amino acids
belonging to proteins outside PG are found in minor
quantities in CW preparations. The results obtained
indicate the occurrence of PG amino acids as follows:
E. aerofaciens ATCC 25986: alanine, glutamic acid,
lysine, aspartic acid; E. aerofaciens ATCC 35085: alanine, glutamic acid, ornithine, aspartic acid; E. limosum
ATCC 8486: alanine, glutamic acid, lysine, ornithine,
serine; E. alactolyticum ATCC 17927: alanine, glutamic
acid, diaminopimelic acid. For PG amino acids the
results are also calculated in micromoles and as molar
ratios, revealing tentative PG types ( Table 2). The total
amount of PG amino acids was 25.5% of dry weight in
E. aerofaciens ATCC 25986, 30.1% in E. aerofaciens
ATCC 35085, 36.1% in E. limosum, and 28.9% in
E. alactolyticum. The amounts of protein amino acids
were <2.2, <4.9, 12.6 and 9.0%, respectively.
Host’s immune status
To study the T cell function of the rats injected with
different preparations of Eubacterium CW, Con A
response of spleen lymphocytes taken at 2 weeks after
the CW injection was determined. Cells from the rats
injected with CW of either strain of E. aerofaciens did
not at all respond to Con A, whereas spleen cells from
the rats injected with CW of E. limosum or E. alactolyticum or with PBS responded in a normal way ( Table 3).
The lack of Con A responses cannot be explained by
changes in the T cell populations, since FACScan analysis using six different markers did not reveal any
differences in the splenocytes between the four groups
of rats ( Table 4). The markers analysed for this purpose
were Pan-T, abTCR, cdTCR, CD4 and T cell activation
markers CD25 and OX-40. The monoclonal antibody
used for staining CD4 (OX-35) also reacts with monocytes and macrophages, resulting in values of CD4+
cells exceeding the Pan-T numbers.
To determine whether an effect on B cell functions
could explain the different arthritogenicities of different
CWs, we studied the serum antibody response against
the homologous (injected) CW. It is apparent that the
results obtained do not reveal any explanation for
Eubacterium cell wall-induced arthritis
279
T 2. Results from amino acid analysis, with tentative PG types
Bacterial strain
Amino acids ( mmol/mg dry weight of CW, molar ratio)
E. aerofaciens ATCC 25986
Ala
0.56
1.4
Ala
0.28
0.3
Ser
0.32
0.6
Ala
0.86
0.7
E. aerofaciens ATCC 35085
E. limosum ATCC 8486
E. alactolyticum ATCC 17927
Glu
0.41
1
Glu
0.85
1
Glu
0.54
1
Glu
1.18
1
Lys
0.66
1.6
Orn
0.56
0.7
Orn
0.72
1.3
Dap
0.21
0.2
Asp
0.36
0.9
Asp
0.43
0.5
Ala
0.40
0.7
Suggested PG type
A4a
A4b
Lys
0.62
1.1
B2a
A1c
Abbreviations as in Fig. 1.
T 3. In vitro Con A response of the splenocytes at 2 weeks after injection of bacterial CW preparations
CW preparation
E. aerofaciens ATCC 25986
E. aerofaciens ATCC 35085
E. limosum ATCC 8486
E. alactolyticum ATCC 17927
PBS
Stimulation indexa
c.p.m.b
<1 (5)
<1 (5)
36.3 ± 33.3 (5)
22.7 ± 25.4 (5)
27.4 ± 19.4 (3)
<0 (5)
<0 (5)
34 973.6 ± 34 178.2 (5)
29 830.0 ± 34 438.2 (5)
17 545.9 ± 13 970.3 (3)
aMean ± standard deviation (..) (n).
bExperimental minus background; mean ± .. (n).
T 4. Ex vivo phenotype analysis of splenocytes at 2 weeks after injection of different bacterial CW preparations
Cell type
Pan-T +
abTCR+
cdTCR+
CD4+
CD25 + /Pan-T +
OX-40 + /CD4+
E. aerofaciens
ATCC 25986
E. aerofaciens
ATCC 35085
E. limosum
ATCC 8486
E. alactolyticum
ATCC 17927
PBS
48.6 ± 4.9a
45.1 ± 4.7
2.0 ± 0.5
53.0 ± 3.3
5.0 ± 1.6
1.7 ± 0.9
43.7 ± 1.9
39.5 ± 2.4
1.9 ± 0.5
48.8 ± 3.2
3.9 ± 0.8
1.9 ± 0.7
46.2 ± 1.1
42.3 ± 2.5
2.1 ± 0.6
52.0 ± 2.4
4.0 ± 0.4
1.6 ± 0.4
48.1 ± 3.5
44.5 ± 4.5
2.1 ± 0.1
52.8 ± 2.7
4.0 ± 0.9
1.6 ± 0.6
49.4 ± 2.3
46.6 ± 2.4
2.0 ± 0.2
53.0 ± 5.4
3.8 ± 0.8
1.3 ± 0.5
aPercentage of positive cells; mean ± .. for five rats (three in PBS group).
T 5. Serum antibodies at 2 weeks after injection of different bacterial CW preparations; homologous CW was used as antigen
Isotype
E. aerofaciens
ATCC 25986
E. aerofaciens
ATCC 35085
E. limosum
ATCC 8486
E. alactolyticum
ATCC 17927
IgG
IgM
IgA
0.36 ± 0.07b
1.51 ± 0.79
1.50 ± 0.68
0.48 ± 0.09
1.15 ± 0.57
1.01 ± 0.30
0.30 ± 0.10
0.58 ± 0.32
0.62 ± 0.20
2.12 ± 0.38
1.82 ± 0.48
1.79 ± 0.42
PBSa
0.28 ± 0.03
0.37 ± 0.10
0.39 ± 0.08
aAnalysed with E. aerofaciens ATCC 25986 CW as antigen.
bOptical density (OD); mean ± .. for five rats (10 rats in E. aerofaciens ATCC 25986 group, eight in PBS group).
the varying arthritogenicity ( Table 5). The most
vigorous antibody response against the injected CW was
observed in the rats injected with the non-arthritogenic
E. alactolyticum CW, particularly in IgG class antibodies
which were quite low in the other groups. It is also
noteworthy that the other non-arthritogenic CW
(E. limosum) did not induce an antibody response to the
same magnitude as did CWs of E. alactolyticum or of
the two E. aerofaciens strains. As a control, we also
determined antibody responses against an irrelevant
antigen to find out whether the experimental groups
would behave differently in this respect. As such an
antigen we used an extract of food pellets, since all
animals were exposed to this in exactly the same way.
Responses against the food antigen and the bacterial
CW preparations cannot be quantitatively compared,
but the results ( Table 6) clearly reveal that the four
experimental groups did not differ in their response
against the unrelated antigens. It is evident, however,
that the CW injection stimulated antibody responses
against food antigens over the level observed in the rats
injected with PBS alone ( Table 6). On the basis of these
X. Zhang et al.
280
T 6. Serum antibodies 2 weeks after injection of different bacterial CW preparations; extract of rat food pellets was used as antigen
Isotype
E. aerofaciens
ATCC 25986
E. aerofaciens
ATCC 35085
E. limosum
ATCC 8486
E. alactolyticum
ATCC 17927
IgG
IgM
IgA
0.47 ± 0.11b
1.55 ± 0.29
1.16 ± 0.18
0.48 ± 0.04
1.27 ± 0.13
0.92 ± 0.08
0.54 ± 0.15
1.26 ± 0.23
1.01 ± 0.14
0.54 ± 0.09
1.36 ± 0.23
1.07 ± 0.14
PBSa
0.42 ± 0.06
0.83 ± 0.10
0.68 ± 0.08
aAnalysed with E. aerofaciens ATCC 25986 CW as antigen.
bOptical density (OD); mean ± .. for five rats (10 rats in E. aerofaciens ATCC 25986 group, eight in PBS group).
findings, including responses against the CW, we concluded that antibody responses do not provide any
explanation for the difference in arthritogenicity between
the eubacterial strains.
Discussion
Several authors have suggested that bacterial CW fragments derived from the intestinal flora may pass the
bowel wall, distribute within the body [9, 10, 34, 35]
and cause a local or systemic reaction leading to arthritis
[36–38]. Studies by Severijnen et al. [22] have shown
that i.p. injection of CW of E. aerofaciens, a main
resident in the human intestinal flora, induces a chronic
arthritis in the rat. In the present study, two strains of
E. aerofaciens were chosen to represent arthritogenic
strains, and E. limosum and E. alactolyticum to represent
non-arthritogenic bacterial strains. To our surprise, the
CW isolated from E. aerofaciens ATCC 35085 induced
only acute arthritis, whereas the other E. aerofaciens
strain (ATCC 25986) induced both acute and chronic
arthritis, as expected. These two strains are 100% identical by the 16S rDNA analysis, but they have slightly
different biochemical and CFA profiles, indicating that
they are different clones of the same species. Therefore,
they provide a challenging opportunity to study what
determines the arthritogenicity of the bacterial CW.
The chemical composition of complex biological materials, including bacterial CWs, can be analysed qualitatively and quantitatively by GC–MS. Relying on this
unique methodology and on the results published previously, we here present putative structures of PGs in the
four CW preparations used (Fig. 3). PG of the arthritogenic E. aerofaciens (ATCC 25986) contains alanine,
glutamine/glutamic acid, lysine and aspartic acid; most
probably it represents PG subgroup A4a having
-aspartic acid in the interpeptide bridge and -lysine
in position 3 of the stem peptide. The non-arthritogenic
E. aerofaciens (ATCC 35085) has the same amino acids
except that lysine is replaced by ornithine. Therefore,
its structure most likely represents subgroup A4b having
-aspartic acid in the interpeptide bridge and -ornithine
in position 3 of the stem peptide. One must admit that
the molar ratios observed ( Table 2) do not exactly
support the quantitative ratios required for the subgroup
A4b. However, the molar ratios obtained can be
regarded only as rough guidelines, particularly since the
CW and not the purified PG was used in the analyses.
PG of E. limosum contains alanine, glutamic acid, lysine,
F. 3. Putative structures and types of PGs used, as suggested
by the present results. The figures indicate position in the stem
peptide. In E. aerofaciens, the presence of -Ala in position 4
(in parentheses) is uncertain. Both strains of E. aerofaciens,
and E. alactolyticum have PG of group A with a cross-linkage
between positions 3 and 4. Eubacterium limosum PG belongs
to group B with a cross-linkage between positions 2 and 4. In
subgroup A4a the interpeptide bridge contains a dicarboxylic
amino acid (-Asp) with -Lys in position 3 of the stem
peptide. In subgroup A4b the interpeptide bridge contains a
dicarboxylic amino acid (-Asp) with -Orn in position 3 of
the stem peptide. Subgroup A1c is without an interpeptide
bridge, with Dap in position 3 of the stem peptide. In subgroup
B2a the interpeptide bridge contains a -diamino acid (-Lys)
with -Orn in position 3 of the stem peptide. Asp, aspartic
acid; other abbreviations as in Fig. 1.
Eubacterium cell wall-induced arthritis
ornithine and serine. The composition agrees with the
E. limosum PG structure of subgroup B2a suggested by
Guinand et al. [39]. This subgroup contains -lysine in
the interpeptide bridge and -ornithine in position 3 of
the stem peptide. Finally, PG of E. alactolyticum contains only alanine, glutamic acid and diaminopimelic
acid. Its structure has previously been described by
Severin et al. [40] to belong to subgroup A1c, without
an interpeptide bridge and with diaminopimelic acid in
position 3 of the stem peptide; our findings are in
concert with this ( Fig. 3).
Does the chemical structure of PG explain why only
one of the four CW preparations used induces chronic
arthritis in the rat? Two types of observation might help
here. First, PGs of group A have been reported to be
significantly more powerful immunostimulants than
those of group B [20, 41]. Second, lysine in position 3
of the stem peptide is known to contribute to the
phlogistic capacity of Gram-positive CWs [23]. On this
basis, it would be understandable that PG of group B
from E. limosum as well as group A PGs from
E. alactolyticum and E. aerofaciens ATCC 35085, all
lacking lysine in position 3, do not induce chronic
arthritis. It will be of interest to see, by studying other
bacterial species and strains, whether such a presence of
lysine is decisive for arthritogenicity. It remains possible
that other factors are also required. For example, a
further difference between the two strains of E. aerofaciens is the higher PS content (49.4%) in the arthritogenic
strain when compared with that in the non-arthritogenic
strain (27.5%); the PG–PS complex is known to be
resistant to mammalian enzymes, and the high carbohydrate content may contribute to prolonged tissue
persistence [11].
If the chemical differences explain how only PG of
E. aerofaciens ATCC 25986 caused chronic arthritis, a
question remains why a significant development of acute
arthritis was seen after injection of PG from E. aerofaciens ATCC 35085 but not after injection of PG from
E. limosum or E. alactolyticum. Three different observations may be important in this regard. First, the total
amount of protein amino acids was lower (<2.2% and
<4.9% of dry weight) in CWs of E. aerofaciens strains
than in those of E. limosum (12.6%) or E. alactolyticum
(9.0%). Second, CWs of both E. aerofaciens strains lack
N-acetylgalactosamine ( Table 1). Third, these two
strains were immunosuppressive, when tested by Con A
response at 2 weeks after the CW injection, at a stage
when signs of the acute arthritis had already subsided
and only the chronic arthritis was present (Fig. 2). Such
an immunosuppression was not observed after injection
of CW from E. limosum or E. alactolyticum. It is
apparent that the same cellular mechanisms are not
involved in the arthritis induction and Con A response.
The importance of these three types of observation
regarding the capacity of CWs to induce acute arthritis
remains at the present unanswered. It is of interest that
the presence of MDP within the PG structure is not
sufficient for induction of acute arthritis, since our strain
281
of E. alactolyticum has MDP and yet no acute arthritis
was observed.
We also studied proliferative lymphocyte responses
against the CW preparations used. However, they
proved to have extremely weak if any stimulatory capacity (data not shown), without any difference between
the four bacterial strains. Likewise, no significant, meaningful differences between the four groups of rats were
observed in the FACScan analysis with six T cell markers
or in antibody responses.
Taken together, the present results suggest that the
chemical structure of the bacterial CW is decisive in
determining arthritogenicity/non-arthritogenicity. Our
results agree with the view that the presence of lysine in
position 3 of the PG stem peptide contributes to the
phlogistic capacity [23]. How this contribution is mediated and what else is required cannot be answered on
the basis of the present experiments. It remains to be
seen, for example, how the presence of lysine in the
critical position affects resistance to enzymatic degradation and tissue persistence known to be important for
the development of chronic arthritis [4, 9–12]. Likewise,
the questions of how and why an i.p. injection of a
phlogistic compound leads predominantly to arthritis
remain open.
Acknowledgements
We gratefully thank Leena Kivistö, Marja-Riitta
Teräsjärvi, and Heli Niittymäki for excellent technical
assistance, Jari Jalava for 16S rDNA sequence analysis,
and Hannele Jousimies-Somer for biochemical reactions
and CFA profiles of the four Eubacterium strains. We
also thank Yong Zhang for help in evaluating the
arthritis, and Erkki Eerola, Janne Komi and Jaakko
Uksila for their kind advice and discussions. This work
was supported by EVO of Turku University Central
Hospital.
References
1. Cromartie WJ, Craddock JG, Schwab JH, Anderle SK,
Yang CH. Arthritis in rats after systemic injection
of streptococcal cells or cell walls. J Exp Med
1977;146:1585–602.
2. Severijnen AJ, Hazenberg MP, van de Merwe JP.
Induction of chronic arthritis in rats by cell wall fragments
of anaerobic coccoid rods isolated from the faecal flora of
patients with Crohn’s disease. Digestion 1988;39:118–25.
3. Severijnen AJ, van Kleef R, Hazenberg MP, van de Merwe
JP. Cell wall fragments from major residents of the human
intestinal flora induce chronic arthritis in rats. J Rheumatol
1989;16:1061–8.
4. Šimelyte E, Rimpiläinen M, Rantakokko K et al. Tissue
distribution and persistence of arthritogenic and nonarthritogenic Eubacterium cell walls. Clin Exp Rheumatol
1999;17:281–8.
5. Stimpson SA, Brown RR, Anderle SK et al. Arthropathic
properties of cell wall polymers from normal flora bacteria.
Infect Immun 1986;51:240–9.
6. Severijnen AJ, van Kleef R, Grandia AA, van der Kwast
TH, Hazenberg MP. Histology of joint inflammation
282
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
X. Zhang et al.
induced in rats by cell wall fragments of the anaerobic
intestinal bacterium Eubacterium aerofaciens. Rheumatol
Int 1991;11:203–8.
Kool J, Gerrits Boeye MY, Severijnen AJ, Hazenberg
MP. Immunohistology of joint inflammation induced in
rats by cell wall fragments of Eubacterium aerofaciens.
Scand J Immunol 1992;36:497–506.
Klasen IS, Kool J, Melief MJ et al. Arthritis by autoreactive T cell lines obtained from rats after injection of
intestinal bacterial cell wall fragments. Cell Immunol
1992;139:455–67.
Hoijer MA, Melief MJ, van Helden Meeuwsen CG,
Eulderink F, Hazenberg MP. Detection of muramic acid
in a carbohydrate fraction of human spleen. Infect Immun
1995;63:1652–7.
Melief MJ, Hoijer MA, Van Paassen HC, Hazenberg MP.
Presence of bacterial flora-derived antigen in synovial
tissue macrophages and dendritic cells. Br J Rheumatol
1995;34:1112–6.
Schwab JH. Phlogistic properties of peptidoglycan–polysaccharide polymers from cell walls of pathogenic and
normal-flora bacteria which colonize humans. Infect
Immun 1993;61:4535–9.
Stimpson SA, Esser RE, Cromartie WJ, Schwab JH.
Comparison of in vivo degradation of 125I-labeled peptidoglycan–polysaccharide fragments from group A and
group D streptococci. Infect Immun 1986;52:390–6.
Stewart-Tull DES. The immunological activities of
bacterial peptidoglycans. Annu Rev Microbiol 1980;34:
311–40.
Johannsen L. Biological properties of bacterial peptidoglycan. APMIS 1993;101:337–44.
Hoijer MA, Melief MJ, Debets R, Hazenberg MP.
Inflammatory properties of peptidoglycan are decreased
after degradation by human N-acetylmuramyl--alanine
amidase. Eur Cytokine Netw 1997;8:375–81.
Hoijer MA, Melief MJ, Calafat J et al. Expression and
intracellular localization of the human N-acetylmuramyl-alanine amidase, a bacterial cell wall-degrading enzyme.
Blood 1997;90:1246–54.
Janusz MJ, Chetty C, Eisenberg RA, Cromartie WJ,
Schwab JH. Treatment of experimental erosive arthritis in
rats by injection of the muralytic enzyme mutanolysin.
J Exp Med 1984;160:1360–74.
Fox A, Brown RR, Anderle SK et al. Arthropathic
properties related to the molecular weight of peptidoglycan–polysaccharide polymers of streptococcal cell walls.
Infect Immun 1982;35:1003–10.
Chetty C, Brown RR, Schwab JH. Edema-producing
activity of group A streptococcal polysaccharide and its
possible role in the pathogenesis of cell wall-induced
polyarthritis. J Exp Med 1983;157:1089–100.
Schleifer KH, Kandler O. Peptidoglycan types of bacterial
cell walls and their taxonomic implications. Bacteriol Rev
1972;36:407–77.
Eerola E, Möttönen T, Hannonen P et al. Intestinal flora
in early rheumatoid arthritis. Br J Rheumatol
1994;33:1030–8.
Severijnen AJ, van Kleef R, Hazenberg MP, van de Merwe
JP. Chronic arthritis induced in rats by cell wall fragments
of Eubacterium species from the human intestinal flora.
Infect Immun 1990;58:523–8.
Burroughs M, Rozdzinski E, Geelen S, Tuomanen E. A
structure–activity relationship for induction of meningeal
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
inflammation by muramyl peptides. J Clin Invest
1993;92:297–302.
Willems A, Collins MD. Phylogenetic relationships of the
genera Acetobacterium and Eubacterium sensu stricto and
reclassification of Eubacterium alactolyticum as
Pseudoramibacter alactolyticus gen. nov., comb. nov. Int
J Syst Bacteriol 1996;46:1083–7.
Jalava J, Mäntymaa ML, Ekblad U et al. Bacterial 16S
rDNA polymerase chain reaction in the detection of intraamniotic infection. Br J Obstet Gynaecol 1996;103:664–9.
Hillier
SL,
Moncla
BJ.
Peptostreptococcus,
Propionibacterium, Eubacterium, and other nonsporeforming anaerobic gram-positive bacteria. In: Murray PR,
Baron EJ, Pfaller MA, Tenover FC, Yolken RH, eds.
Manual of clinical microbiology, 6th edn. Washington,
DC: ASM Press, 1995:587–602.
Eerola E, Lehtonen OP. Optimal data processing procedure for automatic bacterial identification by gas-liquid
chromatography of cellular fatty acids. J Clin Microbiol
1988;26:1745–53.
Rosenthal RS, Dziarski R. Isolation of peptidoglycan and
soluble peptidoglycan fragments. Methods Enzymol
1994;235:253–85.
Gilbart J, Harrison J, Parks C, Fox A. Analysis of the
amino acid and sugar composition of streptococcal cell
walls by gas chromatography–mass spectrometry.
J Chromatogr 1988;441:323–33.
Zhang Y, Gripenberg Lerche C, Soderstrom KO, Toivanen
A, Toivanen P. Antibiotic prophylaxis and treatment of
reactive arthritis. Lessons from an animal model. Arthritis
Rheum 1996;39:1238–43.
Gripenberg Lerche C, Skurnik M, Toivanen P. Role of
YadA-mediated collagen binding in arthritogenicity of
Yersinia enterocolitica serotype O:8: experimental studies
with rats. Infect Immun 1995;63:3222–6.
Zhang Y, Toivanen A, Toivanen P. Experimental Yersiniatriggered reactive arthritis: effect of a 3-week course of
ciprofloxacin. Br J Rheumatol 1997;36:541–6.
Lehman TJ, Allen JB, Plotz PH, Wilder RL. Bacterial cell
wall composition, lysozyme resistance, and the induction
of chronic arthritis in rats. Rheumatol Int 1985;5:163–7.
Lehtonen L, Eerola E, Oksman P, Toivanen P. Muramic
acid in peripheral blood leukocytes of healthy human
subjects. J Infect Dis 1995;171:1060–4.
Lehtonen L, Eerola E, Toivanen P. Muramic acid in
human peripheral blood leucocytes in different age groups.
Eur J Clin Invest 1997;27:791–2.
Hazenberg MP, Klasen IS, Kool J, Ruseler van Embden
JG, Severijnen AJ. Are intestinal bacteria involved in the
etiology of rheumatoid arthritis? APMIS 1992;100:1–9.
Lehtonen L, Kortekangas P, Oksman P, Eerola E, Aro
H, Toivanen A. Synovial fluid muramic acid in acute
inflammatory arthritis. Br J Rheumatol 1994;33:1127–30.
Sartor RB. Role of the enteric microflora in the pathogenesis of intestinal inflammation and arthritis. Aliment
Pharmacol Ther 1997;11(Suppl. 3):17–23.
Guinand M, Ghuysen JM, Schleifer KH, Kandler O. The
peptidoglycan in walls of Butyribacterium rettgeri.
Biochemistry 1969;8:200–7.
Severin AI, Kokeguchi S, Kato K. Chemical composition
of Eubacterium alactolyticum cell wall peptidoglycan. Arch
Microbiol 1989;151:348–52.
Kotani S, Tsujimoto M, Koga T, Nagao S, Tanaka A,
Kawata S. Chemical structure and biological activity
relationship of bacterial cell walls and muramyl peptides.
Fed Proc 1986;45:2534–40.