Content uploaded by Raphaela Goldbach-Mansky
Author content
All content in this area was uploaded by Raphaela Goldbach-Mansky on May 09, 2018
Content may be subject to copyright.
ARTHRITIS & RHEUMATISM
Vol. 44, No. 7, July 2001, pp 1689–1697
© 2001, American College of Rheumatology
Published by Wiley-Liss, Inc.
Chromosomal DNA From a Variety of
Bacterial Species Is Present in Synovial Tissue From
Patients With Various Forms of Arthritis
Herve´C.Ge´rard,
1
Zhao Wang,
1
Geng Feng Wang,
2
Hani El-Gabalawy,
3
Rafaela Goldbach-Mansky,
3
Yong Li,
2
Warqaa Majeed,
2
Haidi Zhang,
2
Ngayin Ngai,
2
Alan P. Hudson,
1
and H. Ralph Schumacher
2
Objective. We and others have reported the pres-
ence of Chlamydia and other bacterial species in joint
specimens from patients with reactive arthritis (ReA).
The present study was conducted to investigate whether
bacteria other than those specified by diagnostic criteria
for ReA could be identified in synovial fluid (SF) or
tissue from patients with various arthritides, and
whether the presence of such organisms corresponds to
particular clinical characteristics in any patient set or
subset.
Methods. DNA in synovial biopsy samples and SF
obtained from 237 patients with various arthritides,
including ReA, rheumatoid arthritis, and undifferenti-
ated oligoarthritis, was assayed by polymerase chain
reaction (PCR) using “panbacterial” primers; we chose
only samples known to be PCR negative for Chlamydia,
Borrelia, and Mycoplasma species. PCR products were
cloned, and cloned amplicons from each sample were
sequenced; DNA sequences were compared against all
others in GenBank for identification of bacterial species
involved.
Results. Ten percent of patient samples were PCR
positive in panbacterial screening assays. Bacterial species
identified belonged to the genera Neisseria,Acinetobacter,
Moraxella,Salmonella,Pseudomonas, and others. Thirty-
five percent of PCR-positive patients showed the presence
of DNA from more than a single bacterial species in
synovium; overall, however, we could identify no clear
relationship between specific single or multiple bacterial
species in the synovium and any general clinical charac-
teristics of any individual or group of patients.
Conclusion. This analysis provides the first sys-
tematic attempt to relate bacterial nucleic acids in the
synovium to clinical characteristics, joint findings, and
outcomes. Many patients with arthritis have bacterial
DNA in the joint, and, in some cases, DNA from more
than a single species is present. However, except for 1
case of a control patient with staphylococcal septic
arthritis, it is not clear from the present study whether
the synovial presence of such organisms is related to
disease pathogenesis or evolution in any or all cases.
Reactive arthritis (ReA) is an inflammatory joint
disease associated with prior infection by any of a
number of specific bacterial pathogens, including Chla-
mydia trachomatis and various species of the genera
Salmonella,Yersinia,Campylobacter, and others (1,2).
Indeed, the American College of Rheumatology (ACR)
criteria for diagnosis of this clinical entity require evi-
dence of urogenital or gastrointestinal infections by such
pathogens (3). Interestingly, while the clinical character-
istics associated with Chlamydia-related ReA and those
associated with ReA caused by enteric infection are
congruent, detailed behavior of the bacteria in the
Dr. Ge´rard’s work was supported by NIH grant AR-47186.
Dr. Hudson’s work was supported by the Department of Veterans
Affairs Medical Research Service and by NIH grant AR-42541. Dr.
Schumacher’s work was supported by the Department of Veterans
Affairs Medical Research Service.
1
Herve´C.Ge´rard, PhD, Zhao Wang, MD, Alan P. Hudson,
PhD: Wayne State University School of Medicine, Detroit, Michigan;
2
Geng Feng Wang, MD, Yong Li, MD, Warqaa Majeed, MD, Haidi
Zhang, MD, Ngayin Ngai, MD, H. Ralph Schumacher, MD: University
of Pennsylvania School of Medicine and Department of Veterans
Affairs Medical Center, Philadelphia, Pennsylvania;
3
Hani El-
Gabalawy, MD, Rafaela Goldbach-Mansky, MD: National Institute of
Arthritis and Musculoskeletal and Skin Diseases, NIH, Bethesda,
Maryland.
Drs. Ge´rard and Z. Wang contributed equally to this work.
Address correspondence and reprint requests to Alan P.
Hudson, PhD, Department of Immunology and Microbiology, Wayne
State University School of Medicine, Gordon H. Scott Hall, 540 East
Canfield Avenue, Detroit, MI 48201.
Submitted for publication August 2, 2000; accepted in revised
form March 14, 2001.
1689
synovium seems to be different. For example, we have
shown that C trachomatis in the joint are intact and
metabolically active (4). In contrast, evidence for enteric
bacteria in the joints of affected patients has been based
primarily on the presence of antigens; chromosomal DNA
from these organisms has not been found in synovia in
most reported cases (e.g., see refs. 5–7).
Recent studies have rendered problematic the
ACR requirement for previous urogenital or gastroin-
testinal infections in the diagnosis of ReA. For example,
serologic and other indirect evidence has associated
infection by Chlamydia pneumoniae, a respiratory patho-
gen, with joint disease (8–11). Since the initial reports,
we and others have used polymerase chain reaction
(PCR) analyses to show that chromosomal DNA from
this organism is indeed present in synovial tissue in a
small but significant number of arthritis patients (12–
14); however, we could not identify a clear set of clinical
characteristics associated with synovial C pneumoniae
infection, in partial contrast to results from another
study (13).
Recently, we demonstrated that synovial C pneu-
moniae, like C trachomatis in the same context, is not
only viable but also metabolically active (14). Thus,
because C pneumoniae is known to cause a strong
inflammatory response at its sites of residence, synovial
presence of this pathogen may have to be considered as
a cause of joint inflammation in some patients. More-
over, some groups have reported the presence of Myco-
bacterium species in the joints of arthritis patients (e.g.,
see ref. 15), although it remains to be determined
whether the organism actually causes joint disease.
Borrelia and Tropheryma whippelii are other examples of
bacteria whose DNA has been identified in the joints of
patients, but whose presence would not meet the official
criteria for the diagnosis of ReA.
The demonstration that C pneumoniae, probably
Mycoplasma species, and other (but not ACR-specified)
bacteria are present in synovial fluid (SF) or tissue in
some patients has raised the question of whether other
non-Chlamydia, nonenteric bacteria are frequently
present in the synovium, and, if so, whether such organ-
isms cause or contribute to synovitis. To address this
issue, several laboratories have used broad-range PCR
systems to search for bacterial DNA in synovial materi-
als; most studies targeted SF, and samples from patients
with a range of arthritides were analyzed (e.g., see refs.
16–21). With minor exceptions, these studies indicated
that DNA from many bacterial groups could be found in
the joints of at least some patients. Organisms found
included those belonging to the genera Pseudomonas,
Bortedella,Acinetobacter,Haemophilus,Clostridium, and
others (19–21). In some studies, the broad-range PCR
product generated was of sufficient length to determine
the genus of the bacterium, but not long enough to
identify the species. Usually, only a single PCR product
was sequenced from any patient sample; thus, whether
⬎1 bacterial species was present in the samples studied
remained unresolved. In one study, sequencing of PCR
products suggested polymicrobial synovial infection
(19). In none of these studies was the relationship
between a specific arthritis and the particular infecting
bacterium examined in any detail.
In this study, we continued the search for other
potential bacterial causes of joint disease. Using an
approach similar to that used by other groups, we
confirmed that nonenteric, non-Chlamydia bacterial spe-
cies can be found at reasonably high frequency in
synovial tissues and/or SF of patients with many forms of
arthritis, and we established that synovial presence of
multiple microbial agents is relatively common. We
further provide a systematic analysis of patient clinical
characteristics, joint findings, and outcomes in relation
to the bacterial species identified.
PATIENTS AND METHODS
Patient samples and preparation of nucleic acids for
analyses. From our large library of DNA preparations made
from synovial biopsy samples and SF obtained from arthritis
patients, we selected 237 samples previously defined to be
PCR negative for C trachomatis,C pneumoniae, several Myco-
plasma species, and Borrelia burgdorferi, as determined by
highly specific and sensitive assays targeting chromosomal
DNA sequences from these organisms. Samples originated
from patients attending arthritis clinics at the Department of
Veterans Affairs Medical Center and the University of Penn-
sylvania School of Medicine, Philadelphia, as well as the Early
Arthritis Study Group at the National Institutes of Health,
Bethesda, MD. SF and biopsy samples were obtained by
arthrocentesis and blind needle synovial biopsy (22). The only
criteria for selection of samples for the present analysis were 1)
PCR negativity for certain bacteria as specified above, and 2)
adequate quantity and quality of the DNA preparation for the
proposed analyses.
The samples selected were obtained from patients with
a variety of arthritides, including ReA, undifferentiated oligo-
arthritis, rheumatoid arthritis (RA), osteoarthritis (OA), and
others; one SF-derived DNA preparation from a patient with
septic arthritis was included as a positive control for the
present analyses. Diagnoses were made according to standard
ACR criteria where possible (3); patients who did not fulfill
these criteria sets were diagnosed as having undifferentiated
monoarthritis, oligoarthritis, or polyarthritis. Treatment out-
comes were defined simply as persistent disease or remission;
the latter was defined as no swollen or tender joints at 1-year
1690 GE
´RARD ET AL
followup and normal erythrocyte sedimentation rate and
C-reactive protein level.
Care was always taken during and after obtaining of
patient samples to prevent extraneous environmental contam-
ination. For example, the skin surface was prepared by first
swabbing with Betadine (Purdue-Frederick, Norwalk, CT) and
subsequently with alcohol prior to obtaining the sample;
samples were immediately placed in sterile microfuge tubes,
and the tubes were closed and snap-frozen in liquid N
2
for
transport to the laboratory. We prepared DNA from patient
materials as described previously (4,14), and care was also
taken at this step to prevent contamination. Briefly, each
sample was transferred frozen from its microfuge tube directly
to a sterile Dounce homogenizer containing buffered phenol
and sterile aqueous extraction buffer, all at 65°C. DNA was
prepared from 1 single sample at a time and in isolation in a
biologic hood. The hood was cleaned between each sample
preparation, and several different hoods were used in different
laboratories; none of the laboratories that allowed us to use
their hoods were engaged in bacteriologic research. Following
homogenization in hot phenol, the aqueous layer from each
sample was extracted with phenol–chloroform several times,
and total nucleic acids were recovered from ethanol precipi-
tates. Pure DNA was prepared by treatment of total nucleic
acids with RNase A.
Panbacterial PCR. The broad-range PCR primer sys-
tem we employed to screen synovial DNA preparations has
been used previously (23). Briefly, this is a non-nested assay
system using two primers: 5⬘-GCGTTAATCGGAA-
TTACTGGGCGTAAG-3⬘and 5⬘-GGTTGCGCTCGTT-
GCGGGACTTAACC-3⬘. Cycling parameters were 4 minutes
at 95°C; then 35 cycles of 1 minute at 95°C, 1 minute at 52°C,
and 1 minute at 72°C; then 10 minutes at 72°C. We used a
non-nested system to limit somewhat the sensitivity of our
assays, since our overall intention was to identify nonchlamy-
dial, nonenteric bacterial species present in synovial samples at
some reasonable titer. This primer system is referred to as
“panbacterial,” since it amplifies 16S ribosomal RNA (rRNA)
gene sequences from a number of bacterial species. As designed,
the system amplifies a DNA fragment of ⬃577 bp (depending on
the organism), spanning the region from nucleotides 501 to 1,077
in the standard (Escherichia coli) 16S rRNA gene.
As with other similar systems, the rationale for design
of this broad-range primer system resides in the observation
that prokaryotic 16S rRNA gene sequences have been con-
served over evolution, and those sequences have been espe-
cially well conserved in particular regions that function impor-
tantly in the ribosome (17,24,25). Design of consensus primers
based on these highly conserved DNA sequence regions allows
amplification of rDNA coding sequences from many, although
not all, bacterial rRNA 16S genes (Table 1). We have not
assessed the sensitivity of the panbacterial system with a large
number of nonchlamydial and other bacteria, but for those we
have assessed (C pneumoniae,C trachomatis,E coli), the
sensitivity generally lies in the range of detection of 10–50
bacterial cells (Ge´rard HC, Hudson AP: unpublished observa-
tions). Each panbacterial screening assay reported here was
done in duplicate independently by two different investigators;
for each assay run, water controls (i.e., no added template)
were interspersed among the samples to be analyzed, and
known negative control DNA preparations from synovial ma-
terials were also included. Pure DNA from either E coli or C
trachomatis was used as a positive control for the panbacterial
screening system. Amplification products were analyzed on
ethidium bromide–stained agarose gels.
Cloning and analysis of PCR products. The synovial
DNA preparations indicated above were subjected to amplifi-
cation using the panbacterial primer system as described
above. For those preparations yielding an amplification prod-
uct, that product was cloned in the pGEM-T vector system
(Promega, Madison, WI) in E coli DH5
␣
cells, as specified by
the manufacturer. One initial clone was selected for DNA
sequence determination from each PCR-positive patient sam-
ple; this is referred to hereinafter as the index clone. Sequenc-
ing was done manually in the laboratory of one of the authors
(APH) using the Sequenase 2 system (Amersham Life Sciences,
Cleveland, OH) or at the facilities available at Wayne State
University School of Medicine and the University of Pennsylvania
School of Medicine; both strands were sequenced for each clone.
Bacterial species whose DNA was amplified in these assays were
identified by BLAST search of all DNA sequences in GenBank.
To assess whether the bacterial species identified
above were the only organisms present in each patient sample,
we further analyzed several cloned PCR products from each
sample, in addition to that of the index clone. We first
performed a detailed computer analysis of all possible restric-
tion enzyme cleavage sites for each cloned index PCR product,
based on the sequencing results from each clone. We then
prepared DNA from 4–7 additional clones from each patient
DNA preparation that was PCR positive in the panbacterial
system. On each of these additional clones, we performed an
extensive restriction analysis. We first investigated whether the
clone insert was the same size as that of the cognate index
clone, via restriction analysis as visualized on agarose gels (26).
We then investigated whether 3–5 restriction enzyme cleavage
sites predicted from the sequence of the index clone insert
were present in all other clones from the same patient, and
Table 1. Bacterial genera targeted by panbacterial polymerase chain
reaction primers*
Bacillus
Borrelia
Campylobacter
Escherichia
Listeria
Neisseria
Proteus
Rickettsia
Salmonella
Staphylococcus
Yersinia
Bordetella
Chlamydia
Klebsiella
Mycoplasma
Pseudomonas
Shigella
Serratia
Streptococcus
Other
* Determined by homology search of primer sequences in GenBank.
BACTERIAL DNA IN THE SYNOVIUM 1691
whether 3–4 enzymes predicted not to cut the insert did so. For
all additional clones with insert size different from that of the
relevant index clone, and for clone inserts whose restriction
pattern did not match that of the relevant index clone, the
DNA sequence of the clone insert was determined as above.
Those sequences were then compared with all others in
GenBank, as described above, for identification of bacterial
species.
RESULTS
PCR positivity in the panbacterial assay system.
We analyzed 237 DNA preparations from synovial biopsy
samples (n ⫽232) and some SF samples (n ⫽5) for the
presence of bacterial DNA, using a set of consensus
primers targeting a conserved region of the prokaryotic
16S rRNA gene. All DNA preparations studied were
known to be PCR negative in assays targeting C trachoma-
tis,C pneumoniae,B burgdorferi, and several species of
Mycoplasma. Of the 237 patient DNA preparations ana-
lyzed, 23 (9.7%) were PCR positive. Table 2 presents the
distribution of PCR positivity in these assays among the
patient groups with various diagnoses, and Table 3 pro-
vides a general summary of clinical details for each patient
whose DNA was PCR positive in the panbacterial assay.
With the exception of 1 sample, all PCR-positive prepara-
tions were from synovial tissue. DNA from a control SF
sample (patient C, with known septic arthritis) was also
PCR positive in the assays, as expected.
Bacterial species identified in index clones. We
cloned the panbacterial PCR product from each patient
Table 2. General summary of polymerase chain reaction (PCR)
positivity of patient DNA preparations in panbacterial assays as a
function of diagnosis*
Diagnosis
Total no.
of patients
No. (%) of
patients with
PCR-positive samples
Septic arthritis (control) 1 1 (100)
Behc¸et’s disease 1 1 (100)
Psoriatic arthritis 2 2 (100)
RF⫺polyarthritis 6 6 (100)
OA 61 3 (4.9)
Umono 1 1 (100)
UO 85 1 (1.2)
ReA 36 3 (8.3)
RA 39 4 (10.3)
Unknown† 5 1 (20)
Total 237 23 (9.7)
*RF ⫽rheumatoid factor; OA ⫽osteoarthritis; Umono ⫽undiffer-
entiated monarthritis; UO ⫽undifferentiated oligoarthritis; ReA ⫽
reactive arthritis; RA ⫽rheumatoid arthritis.
† Diagnoses and/or other data were missing for 5 patients.
Table 3. Characteristics of patients with PCR-positive DNA preparations in panbacterial assays*
Patient/age/sex
Working
diagnosis Disease duration Sample type
C/61M Septic arthritis 2 weeks SF
1/36/F RF⫺polyarthritis 0.5 years ST
2/27/M RF⫺polyarthritis 2 years ST
3/28/M RF⫺polyarthritis 3 years ST
4/30/M RF⫺polyarthritis 1 year ST
5/38/F RF⫺polyarthritis 0.75 year ST
6/40/F RF⫺polyarthritis 0.75 year ST
7/49/F ReA 0.1 year ST
8/25/F ReA 2 years ST
9/54/M ReA 4 years ST
10/40/F UO ⬍1 year ST
11/32/M Umono 2.5 years ST
12/46/F RA ⬍1 year ST
13/29/F RA 3 years ST
14/62/M RA ⬍1 year ST
15/57/F RA 0.75 year SF
16/50/M RA 8 years ST
17/33/M Psoriatic arthritis ⬍1 year ST
18/36/F Psoriatic arthritis ⬍1 year ST
19/27/F Behc¸et’s disease 4 years ST
20/78/M OA 2 years ST
21/32/M OA 1 year ST
22/77/M† OA NA ST
23/NA† NA NA ST
*C ⫽control; SF ⫽synovial fluid; ST ⫽synovial tissue; NA ⫽not available (see Table 2 for other
definitions).
† Some data were missing for patients 22 and 23.
1692 GE
´RARD ET AL
sample, determined the DNA sequence of 1 clone insert
(the index clone) from each sample, and compared that
index clone sequence with all DNA sequences in GenBank
to identify the particular bacterial species from which the
PCR product originated. Table 4 shows the results of these
analyses. Unlike findings in some previous studies, the
PCR product generated by the panbacterial primer system
used here was generally of sufficient length (⬃577 bp,
depending upon the particular bacterial species) to define
the infecting organism at the species level. For example,
sequence analysis of the index clone insert from patient C,
the positive control sample, indicated the presence of DNA
from Staphylococcus aureus, the most common organism
associated with septic arthritis. As in previous similar
studies, we found a diverse array of bacterial species to be
present in synovial samples from arthritis patients, includ-
ing Moraxella osloensis,Klebsiella planticola,Neisseria canis,
several species of Pseudomonas,Stenotrophomonas malto-
philaia, and others. No particular species of bacterium
identified in these assays was predominant, although sev-
eral species from the genus Pseudomonas were identified.
Presence of DNA from multiple bacterial species
in some patients. In addition to the index clone, we
prepared DNA from several other clones for each
PCR-positive patient DNA preparation, and we sub-
jected each additional clone to further analysis to deter-
mine whether bacterial species other than that identified
in the index clone were present in the patient sample.
Figure 1 presents a representative additional analysis.
The insert to the index clone for patient 12 was ⬃580 bp
in length (lane 1), and DNA sequencing of the insert
indicated that the amplicon was derived from the 16S
rRNA gene of N canis. The insert sizes for 4 additional
clones from the same amplification were indistinguish-
able from that of the index clone (lanes 2–5), and
restriction of all inserts using the enzymes Eco RI (lanes
6–10) or Cla I (lanes 11–15) also indicated no differ-
ences in comparison with that of the index clone.
Restriction of the clone inserts using the enzyme Bsm I
indicated recognition sites in the index clone, as ex-
pected (lane 16), and in additional clones 2–4 (lanes
18–20); cleavage at those 2 sites generated a DNA
Table 4. Bacterial species identified in index clones and in additional clones*
Patient Working diagnosis Index clone Additional clones†
C Septic arthritis Staphylococcus aureus –
1RF⫺polyarthritis Acinetobacter calcoaceticus Xanthomonas axonopodis,
Stenotrophomonas maltophilaia
2RF⫺polyarthritis Acinetobacter calcoaceticus Pseudomonas putida/borealis‡
3RF⫺polyarthritis Salmonella typhi Salmonella enteritidis or
Escherichia coli
4RF⫺polyarthritis Salmonella enteritidis Same
5RF⫺polyarthritis Pseudomonas putida/borealis‡ Acinetobacter calcoureticus
6RF⫺polyarthritis Klebsiella oxytoca Same
7 ReA Shigella dysenteriae Mycoplasma mycoides or
Mycoplasma capricolum‡
8 ReA Pseudomonas putida/borealis‡ Same
9 ReA Pseudomonas putida/borealis‡ Same
10 UO Klebsiella oxytoca Same
11 Umono Pseudomonas fluorescens Same
12 RA Neisseria canis Pseudomonas putida/borealis‡
13 RA Moraxella osloensis Same
14 RA Salmonella typhi Enterobacter agglomerans,
Klebsiella planticola
15 RA Shigella flexneri Same
16 RA Pseudomonas testosteroni Same
17 Psoriatic arthritis Klebsiella planticola Same
18 Psoriatic arthritis Pseudomonas putida/borealis‡ Same
19 Behc¸et’s disease Enterobacter cloacae Same
20 OA Pseudomonas putida/borealis‡ Same
21 OA Shigella dysenteriae Same
22 OA Pseudomonas hibisciola Stenotrophomonas maltophilaia
23 NA Pseudomonas stutzeri ND
*ND ⫽not determined (see Tables 2 and 3 for other definitions and explanations).
† “Same” indicates that all additional clones were the same organism as the index clone.
‡Pseudomonas putida and Pseudomonas borealis could not be differentiated on the basis of the available
16S ribosomal RNA sequence (see Figure 1). For the same reason, Salmonella enteritidis could not be
distinguished from Escherichia coli, and Mycoplasma mycoides could not be distinguished from Myco-
plasma capricolum.
BACTERIAL DNA IN THE SYNOVIUM 1693
fragment of ⬃100 bp in each plasmid insert. Additional
clone 1 (lane 17) also showed the presence of Bsm I sites
in the insert, but cleavage of that insert with the enzyme
generated an internal restriction fragment of ⬃125 bp,
significantly different from that of the index clone or
additional clones 2–4 for this sample. DNA sequence
analysis of the insert to additional clone 1 from patient
12 showed it to be from the 16S rRNA gene of Pseudo-
monas putida or Pseudomonas borealis; these closely
related species are not distinguishable on the basis of
the DNA sequence of this portion of the 16S rRNA
molecule.
Table 4 presents a summary of index clone
species information and also summarizes the additional
bacterial species that were found in patient samples. In
8 of 23 samples (34.8%), DNA from more than a single
bacterial species was present in our patient group. The
multiple species present in the individual samples were
not always closely related to one another. Moreover,
many of these species are not considered pathogenic by
normal standards. The wide variety of bacterial DNA
detected in our assays suggests that there was not a
systematic laboratory contamination of our samples.
Pseudomonas species are, however, common environ-
mental organisms, and we cannot absolutely rule out the
possibility of some level of sample contamination with
these organisms (but see below).
Clinical features of patients with single and
multiple bacterial DNA in synovial tissue. As mentioned
above, only patient C (the control) had a diagnosis of
septic arthritis, and in this patient we identified S aureus
in the synovium, as expected. Three patients with chro-
mosomal DNA from Klebsiella species in their index
clone inserts included 1 with persistent psoriatic arthritis
(patient 17), 1 with a seronegative polyarthritis (patient
6), and 1 with undifferentiated oligoarthritis, which
resolved (patient 10); 1 patient with RA (patient 14) had
K planticola in an additional clone but not in his index
clone, and this patient had DNA from an Enterobacter
species and from Salmonella typhi as well. Nine patients
had DNA from various Pseudomonas species in their
index clones (patients 5, 8, 9, 11, 16, 18, 20, 22, and 23),
and 2 other patients had DNA from Pseudomonas
species in additional clones (patients 2 and 12). In the
former group, 1 patient had psoriatic arthritis and 1 had
arthritis associated with hydradenitis (patient 11), and
the latter patient had an intensely inflammatory monar-
thritis as well. All these individuals had persistent dis-
ease. One might speculate that the patient with psoriatic
arthritis and the one with hydradenitis might be espe-
cially prone to contamination from skin, with persistence
Figure 1. Representative analysis of an index clone and additional
clones to determine the presence of DNA from single or multiple
bacterial species in patient samples. The DNA preparation analyzed is
from patient 12. DNA was prepared from synovial tissue, and the
analyses performed as described in Patients and Methods and in
Results. Cloning and DNA sequence analysis of the index clone for
patient 12 indicated that the amplicon was derived from the 16S
ribosomal RNA (rRNA) gene of Neisseria canis. Top gel, 100-bp size
standards (lane C
1
); index clone (lane 1) and additional clones 1–4
(lanes 2–5) treated with restriction enzymes Nde I and Sst I to release
clone inserts (note that all inserts are the same size); index clone (lane
6) and additional clones 1–4 (lanes 7–10) treated with restriction
enzymes Nde I and Sst I to release clone inserts, as well as with
restriction enzyme Eco RI, which should find 2 sites internal to the N
canis insert DNA sequence to generate a smaller fragment of 165 bp
(note that all inserts show the smaller fragment of correct and
equivalent size). Bottom gel, 100-bp size standards (lane C
2
); index
clone (lane 11) and additional clones 1–4 (lanes 12–15) treated with
restriction enzymes Nde I and Sst I to release clone inserts, as well as
with restriction enzyme Cla I, which should not have sites within the N
canis insert DNA sequence (note that the site is absent from all insert
DNA sequences); index clone (lane 16) and additional clones 1–4
(lanes 17–20) treated with restriction enzymes Nde I and Sst Ito
release clone inserts, as well as with restriction enzyme Bsm I, which
should find 2 sites internal to the insert to generate a fragment of 100
bp (note that additional clones 2–4 show the 100-bp fragment, while
additional clone 1 shows a fragment of ⬃125 bp). DNA sequence
analysis of the insert to additional clone 1 indicated that it was derived
from the 16S rRNA gene of either Pseudomonas putida or Pseudomo-
nas borealis; these 2 closely related species cannot be distinguished on
the basis of the available DNA sequence.
1694 GE
´RARD ET AL
of bacterial DNA in the inflamed joint. The other
patients with Pseudomonas DNA included those with
RA, OA, ReA, and undifferentiated polyarthritis, and
all these patients did well on followup without any
specific antibiotic therapy.
Three patients had DNA from Shigella species in
their index clones (patients 7, 15, and 21), and 3 other
patients had DNA from Salmonella species in their index
clones (patients 3, 4, and 14); one of these (patient 3)
had DNA from a different Salmonella species in an
additional clone. All of these are difficult to explain as
contaminants. None of these 6 individuals had clinical
septic arthritis. Three patients in this subgroup had no
antecedent gastrointestinal symptoms, and 2 of the
patients with Shigella DNA had no antibodies to the
organism. Each of the patients studied via SF analysis
actually had low SF leukocyte counts in the joint exam-
ined, although 2 had clinical diagnoses of RA. Only 1
patient had arthralgia (patient 4), 1 had mild OA
(patient 21), and 1 had ReA (patient 7). Two patients
with RA had synovitis for several years, and both are
now doing well on followup unassisted by antibiotic
treatment.
Consistent with data from similar studies, we
identified DNA from a number of relatively unfamiliar
organisms (e.g., S maltophilaia, M osloensis), and we
could find no suggestion of clear clinical associations
with the presence of these bacteria, with the possible
exception of patient 19 with Behc¸et’s disease; this was
the only individual identified with Enterobacter cloacae
DNA.
DISCUSSION
Previous studies have demonstrated the presence
of a wide variety of bacterial species in the synovia of
arthritis patients (8–16,19–21), raising the question of
whether the organisms identified might have some asso-
ciation with the patients’ various joint diseases. In the
present study, we assessed DNA preparations from SF
and synovial tissue of 237 arthritis patients known to be
PCR negative for C trachomatis, C pneumoniae,B
burgdorferi, and several Mycoplasma species, and we
showed that ⬃10% of such patients had DNA from
diverse bacteria in the synovium. The bacteria identified
in our analyses include species from the genera Kleb-
siella, Pseudomonas, Salmonella, Shigella, Neisseria, and
Acinetobacter, groups not significantly different from
those identified in other, previous studies; we did iden-
tify species from the genera Moraxella or Stenotrophomo-
nas, which, to our knowledge, have not been identified
before in this context. Interestingly, however, and prob-
ably most important, ⬃35% of patients who were PCR
positive for one bacterial species also showed the pres-
ence of a different, not necessarily related or pathogenic,
bacterial species in synovium. Taken together with those
of other investigators, our observations clearly suggest
that bacteria are rather commonly present in the joint,
but it is still not clear whether such presence is associ-
ated with joint disease (see below).
The proportion of our patients who were PCR
positive for bacterial DNA other than that from C
trachomatis and/or enteric bacteria in synovium was
somewhat lower than the proportions reported by other
groups. For example, in one study using generalized 16S
rRNA–directed primers, 37% of SF samples showed the
presence of bacterial DNA (16), while in another study,
42% of SF and synovial tissue samples assessed were
PCR positive in such assays (19). Our lower PCR-
positivity rate is due in part to our exclusion of patient
samples known to be PCR positive for Chlamydia, etc.,
and in part to the non-nested design of our panbacterial
assay system.
Interestingly, one recent report indicated that
16S rRNA from an extraordinarily large number of
bacterial species could be identified in synovia of both
control and arthritis patient samples (27). Reverse
transcription–PCR with generalized primers, rather than
PCR, was used for screening in that study; however,
there appeared to be little difference in the proportions
of control and arthritis patients with synovial prokary-
otic DNA. A previous study by our group showed that C
trachomatis chromosomal sequences were present in
⬃5% of DNA samples from synovial tissue of normal
control individuals (i.e., individuals with no apparent
joint disease) (28); other studies in which we used
synovial tissues from OA patients as controls also
showed a PCR-positivity rate of ⬃5% for C trachomatis
DNA (29).
Thus, it seems clear that there is usually a low-
level “background” presence of bacteria in synovial
material, and that such organisms do not necessarily
cause synovitis. It remains to be determined, of course,
whether the presence in synovium of bacteria other than
C trachomatis,Salmonella, Yersinia, and others specified
by ACR diagnostic criteria is responsible for synovitis in
at least some patients and, if so, precisely how patho-
genesis is initiated and/or maintained. Along this same
line, we suspect that host genetic factors are critically
important in many, if not most, cases of prokaryote-
induced joint disease or disease exacerbation. Whatever
the answers to these questions, accumulating evidence
BACTERIAL DNA IN THE SYNOVIUM 1695
indicates that a variety of bacterial nucleic acids are
present in joints, and the implication of their presence
needs to be understood. A related issue is whether other
patient tissues and organs harbor nucleic acids from
single or multiple bacterial species and, if not, what this
suggests concerning synovial biology.
Our results indicate a high prevalence of polymi-
crobial agents in the synovium. This observation is consis-
tent with those of at least one previous study (19), and it
emphasizes that our knowledge of the dynamics of human-
microbe ecology is relatively poor. This lack of understand-
ing is especially important in relation to the long-term
clinical consequences of the presence of bacteria (either
single or multiple species) in the joint. In our view, it is
reasonable to argue that, given the presence of certain
as-yet-undefined host genetic factors, synovial bacteria may
either initiate synovial inflammation or, perhaps more
likely, exacerbate inflammation initiated by some other
means. The presence of multiple bacterial species in the
joint also may increase the probability that inflammation or
other pathogenetic processes will ensue in some patients.
Clearly, it will be important to elucidate interactions not
only between bacterium and host, but also between/among
bacterial species and other infectious agents in the synovial
context.
Interestingly, of the sample types studied here from
which nucleic acids were prepared, the overwhelming
majority of PCR-positive samples in the panbacterial as-
says were from synovial tissue rather than SF. Indeed, only
a single SF sample of those assayed was PCR positive, and
this is in contrast to earlier results from other groups, in
which a higher proportion of SF samples were positive in
the generalized PCR assays employed (19,20). The reasons
for this discrepancy are not clear, but they may be related
either to the primer systems used or to the clinical charac-
teristics of the patients from whom samples were obtained.
We do not know whether any, several, or all of the bacterial
species identified in our patient samples can and do
establish long-term joint residence, nor do we know
whether the organisms identified in our analysis were
metabolically active when the samples were obtained.
Importantly, we also do not understand how microbial
populations in the synovium change over time, or, in the
case of polymicrobial populations, how the composition of
the population affects interaction with the host. Elucida-
tion of these and other related issues is critical and will
require further study.
While organisms belonging to most of the genera
identified in our panbacterial screening analyses have
been found in studies by other groups, some species
identified here in synovial tissue deserve comment. For
example, we were not surprised to find Shigella dysente-
riae, Shigella flexneri, Salmonella enteritidis, and Klebsiella
oxytoca in the samples analyzed, but we were surprised
to find S typhi in 2 patient samples. We were also
surprised to find K planticola and S maltophilaia; these
are both environmental organisms that have not been
considered significant pathogens in humans. However,
both apparently can cause wound infections under some
circumstances (30,31).
Many Pseudomonas species are ubiquitous in the
environment, but the spectrum of bacteria identified in
our analyses probably does not represent sample con-
tamination to any meaningful degree, given the ex-
tremely stringent precautions taken in preparation and
analyses (12,32); however, such an eventuality can never
be excluded completely. It is noteworthy in this respect
that we did not identify Bacillus subtilis in any of our
samples; the presence of this organism was considered
an environmental contaminant in one earlier study (20).
Importantly, recent studies from our group have indi-
cated that plugs of skin are commonly introduced into
the joint with needle entry (ref. 33 and Schumacher HR:
unpublished observations). We propose that future stud-
ies to evaluate the presence of bacterial nucleic acids in
joint materials should be supplemented with morpho-
logic study of the synovium, in situ hybridization (34), or
electron microscopy (35) to confirm the presence of
organisms at relevant tissue sites.
Given the diagnoses of those patients having
single or multiple bacterial species in synovium, we
could not conclude that any specific joint disease was
linked to the presence of bacteria. Moreover, detailed
examination of patient histories proved to be unenlight-
ening, since we could find no suite of clinical character-
istics that was unequivocally associated with the pres-
ence of ⱖ1 bacterial species in the joint, although it is of
interest that both individuals with psoriatic arthritis and
all 6 patients with rheumatoid factor–negative polyar-
thritis were panbacteria positive. We note, however, that
a high percentage of the patients studied were in disease
remission at followup. This lack of clinical pattern is not
surprising, since the organisms identified are diverse in
their biology and pathogenic potential. Indeed, in a
previously published study, we could identify no set of
clinical characteristics associated with the presence in
synovium of C pneumoniae (12); in that study, we
attributed this lack of associated characteristics to our
lack of understanding of the biology of this organism,
with its apparent involvement in clinical entities as
diverse as respiratory tract infections and atheroscle-
rosis.
1696 GE
´RARD ET AL
We ascribe our failure to define a set (or multiple
sets specific to various groups of bacteria) of disease
attributes in the present study to our general lack of
understanding of human-microbe ecology, and to our
ignorance of much prokaryotic biology, discussed earlier
in this section. More study will be required to remedy
these shortcomings.
REFERENCES
1. Schumacher HR. Reactive arthritis. Rheum Dis Clin North Am
1998;24:261–73.
2. Ko¨hler L, Zeidler H, Hudson AP. Etiologic agents in reactive
arthritis: their molecular biology and phagocyte-host interactions.
Baillieres Clin Rheumatol 1998;12:589–609.
3. Schumacher HR, Klippel JH, Koopman WJ, editors. Primer on the
rheumatic diseases. 10th ed. Atlanta: Arthritis Foundation; 1993.
4. Ge´rard HC, Branigan PJ, Schumacher HR, Hudson AP. Synovial
Chlamydia trachomatis in patients with reactive arthritis/Reiter’s
syndrome are viable but show aberrant gene expression. J Rheu-
matol 1998;25:734–42.
5. Granfors K, Jalkanen S, Lindberg AA, Maki-Ikola O, von Essen
R, Lahesmaa-Rantala R, et al. Salmonella lipopolysaccharide in
synovial cells from patients with reactive arthritis. Lancet 1990;
335:685–8.
6. Granfors K, Jalkanen S, von Essen R, Lahesmaa-Rantala R,
Isomaki O, Pekkola-Heino K, et al. Yersinia antigens in synovial
fluid cells from patients with reactive arthritis. N Engl J Med
1989;320:216–21.
7. Merilahti-Palo R, So¨derstro¨m KO, Lahesmaa-Rantala R, Gran-
fors K, Toivanen A. Bacterial antigens in synovial biopsy speci-
mens in Yersinia-triggered reactive arthritis. Ann Rheum Dis
1991;50:87–90.
8. Gran JT, Hjetland R, Andraessen AH. Pneumonia, myocarditis,
and reactive arthritis due to Chlamydia pneumoniae. Scand
J Rheumatol 1993;22:43–4.
9. Braun J, Laitko S, Treharne J, Eggens U, Wu P, Distler A, et al.
Chlamydia pneumoniae—a new causative agent of reactive arthritis
and undifferentiated oligoarthritis. Ann Rheum Dis 1994;53:100–5.
10. Beaudreuil J, Hayem G, Meyer O, Khan MF. Reactive arthritis
ascribed to Chlamydia pneumoniae. Rev Rhum Engl Ed 1995;3:
222–4.
11. Wilkinson NZ, Kingsley GH, Sieper J, Braun J, Ward ME. Lack of
correlation between the detection of Chlamydia trachomatis DNA
in synovial fluid from patients with a range of rheumatic diseases
and the presence of an antichlamydial immune response. Arthritis
Rheum 1998;41:845–54.
12. Schumacher HR Jr, Ge´rard HC, Arayssi TK, Pando JA, Branigan
PJ, Saaibi DL, et al. Lower prevalence of Chlamydia pneumoniae
DNA compared with Chlamydia trachomatis DNA in synovial
tissue of arthritis patients. Arthritis Rheum 1999;42:1889–93.
13. Hannu T, Puolakkainen M, Leirisalo-Repo M. Chlamydia pneu-
moniae as a triggering agent in reactive arthritis. Rheumatology
(Oxford) 1999;38:411–4.
14. Ge´rard HC, Schumacher HR, El-Gabalawy H, Goldbach-Mansky
R, Hudson AP. Chlamydia pneumoniae present in the human
synovium are viable and metabolically active. Microb Pathogen
2000;29:17–24.
15. Schaeverbeke T, Gilroy CB, Bebear C, Dehais J, Taylor-Robinson
D. Mycoplasma fermentans, but not M penetrans, detected by
PCR assays in synovium from patients with rheumatoid arthritis
and other rheumatic disorders. J Clin Pathol 1996;49:824–8.
16. Wilkinson NZ, Kingsley GH, Jones HW, Sieper J, Braun J, Ward
ME. Detection of DNA from a range of bacterial species in the
joints of patients with a variety of arthritides using a nested,
broad-range polymerase chain reaction. Rheumatology (Oxford)
1999;38:260–6.
17. Relman DA. The identification of uncultured microbial patho-
gens. J Infect Dis 1993;168:1–8.
18. Relman DA. The search for unrecognized pathogens. Science
1999;284:1308–10.
19. Wilbrink B, van der Heijden IM, Schouls LM, van Embden JDA,
Hazes JMW, Breedveld FC, et al. Detection of bacterial DNA in
joint samples from patients with undifferentiated arthritis and
reactive arthritis, using polymerase chain reaction with universal
16S ribosomal RNA primers. Arthritis Rheum 1998;41:535–43.
20. Gray J, Marsh PJ, Walker DJ. A search for bacterial DNA in RA
synovial fluids using polymerase chain reaction. Br J Rheumatol
1994;33:997–9.
21. Li F, Bulbul R, Schumacher HR, Kieber-Emmons T, Callegari PE,
von Feldt JM, et al. Molecular detection of bacterial DNA in
venereal-associated arthritis. Arthritis Rheum 1996;39:950–8.
22. Schumacher HR, Kulka JB. Needle biopsy of the synovial mem-
brane: experience with the Parker-Pearson technique. N Engl
J Med 1972;286:416–9.
23. Gerard H, Hudson A, Williams W, Yarboro C, Klippel J, Branigan P,
et al. Screening synovial fluid and tissue using PCR for pan bacterial
16S rRNA [abstract]. Arthritis Rheum 1997;40 Suppl 9:S227.
24. Olsen GJ, Woese CR. Ribosomal RNA: a key to phylogeny.
FASEB J 1993;1:113–23.
25. Gutell RR, Larsen N, Woese CR. Lesson from an evolving rRNA:
16S and 23S structures from a comparative perspective. Microbiol
Rev 1994;58:10–26.
26. Sambrook J, Fritsch EF, Maniatis R. Molecular cloning: a labo-
ratory manual. 2nd ed. Cold Spring Harbor (NY): Cold Spring
Harbor Laboratory Press; 1989.
27. Kempsell KE, Cox CJ, Hurle M, Wong A, Wilkie S, Zanders ED,
et al. Reverse transcription-PCR of bacterial rRNA for detection
and characterization of bacterial species in arthritis synovial tissue.
Infect Immun 2000;68:6012–26.
28. Schumacher HR Jr, Arayssi T, Crane M, Lee J, Gerard H, Hudson
AP, et al. Chlamydia trachomatis nucleic acids can be found in the
synovium of some asymptomatic subjects. Arthritis Rheum 1999;
42:1281–4.
29. Branigan PJ, Ge´rard HC, Hudson AP, Schumacher HR, Pando
J. Comparison of synovial tissue and synovial fluid as the source of
nucleic acids for detection of Chlamydia trachomatis by polymer-
ase chain reaction. Arthritis Rheum 1996;39:1740–6.
30. Krieg NR, Holt JG, editors. Bergey’s manual of systematic bacte-
riology. Baltimore: Williams & Wilkins; 1984.
31. Baron S, editor. Medical microbiology. New York: Churchill
Livingstone; 1991.
32. Balin BJ, Ge´rard HC, Arking EJ, Appelt DM, Branigan PJ,
Abrams JT, et al. Identification and localization of Chlamydia
pneumoniae in the Alzheimer’s brain. Med Microbiol Immunol
(Berl) 1998;187:23–42.
33. Glaser DL, Schildhorn JC, Bartolozzi AR, Dennis R, Li Y, Amato
TR, et al. Do you really know what is on the tip of your needle?
The inadvertent introduction of skin into the joint [abstract].
Arthritis Rheum 2000;43 Suppl 9:S149.
34. Beutler AM, Hudson AP, Whittum-Hudson JA, Salameh WA,
Gerard HC, Branigan PJ, et al. Chlamydia trachomatis can persist in
joint tissue following antibiotic treatment of patients with Reiter’s
syndrome/reactive arthritis. J Clin Rheumatol 1997;3:125–30.
35. Nanagara R, Li F, Beutler A, Hudson A, Schumacher HR Jr.
Alteration of Chlamydia trachomatis biologic behavior in synovial
membranes: suppression of surface antigen production in reactive
arthritis and Reiter’s syndrome. Arthritis Rheum 1995;38:1410–7.
BACTERIAL DNA IN THE SYNOVIUM 1697
