Analysis of bacterial DNA in synovial tissue of Tunisian patients with reactive and undifferentiated arthritis by broad-range PCR, cloning and sequencing

Introduction Bacteria and/or their antigens have been implicated in the pathogenesis of reactive arthritis (ReA). Several studies have reported the presence of bacterial antigens and nucleic acids of bacteria other than those specified by diagnostic criteria for ReA in joint specimens from patients with ReA and various arthritides. The present study was conducted to detect any bacterial DNA and identify bacterial species that are present in the synovial tissue of Tunisian patients with reactive arthritis and undifferentiated arthritis (UA) using PCR, cloning and sequencing. Methods We examined synovial tissue samples from 28 patients: six patients with ReA and nine with UA, and a control group consisting of seven patients with rheumatoid arthritis and six with osteoarthritis (OA). Using broad-range bacterial PCR producing a 1,400-base-pair fragment from the 16S rRNA gene, at least 24 clones were sequenced for each synovial tissue sample. To identify the corresponding bacteria, DNA sequences were compared with sequences from the EMBL (European Molecular Biology Laboratory) database. Results Bacterial DNA was detected in 75% of the 28 synovial tissue samples. DNA from 68 various bacterial species were found in ReA and UA samples, whereas DNA from 12 bacteria were detected in control group samples. Most of the bacterial DNAs detected were from skin or intestinal bacteria. DNA from bacteria known to trigger ReA, such as Shigella flexneri and Shigella sonnei, were detected in ReA and UA samples of synovial tissue and not in control samples. DNA from various bacterial species detected in this study have not previously been found in synovial samples. Conclusion This study is the first to use broad-range PCR targeting the full 16S rRNA gene for detection of bacterial DNA in synovial tissue. We detected DNA from a wide spectrum of bacterial species, including those known to be involved in ReA and others not previously associated with ReA or related arthritis. The pathogenic significance of some of these intrasynovial bacterial DNAs remains unclear.


Introduction
Bacteria are considered to be important in the pathogenesis of several forms of arthritis, including reactive arthritis (ReA) (page number not for citation purposes) ble in the joints [3,4]. Usually, the initial arthritogenic bacterial infection affects the urogenital tract (for example, Chlamydia trachomatis) or the digestive tract (Yersinia, Salmonella or Shigella spp., or Campylobacter jejeuni) [4]. ReA can also follow respiratory tract infections with Chlamydophila pneumoniae [5].
Many cases of ReA are preceded by infections that are asymptomatic [6]; such cases are clinically classified as undifferentiated arthritis (UA) [7,8]. This term describes patients who exhibit arthritis clinically similar to ReA, with high rates of monoarthritis or oligoarthrithis, and predominance of synovitis in the lower limbs. This has led several investigators to suggest a potential link between these two forms of arthritis, and that UA and ReA are overlapping entities. Several groups have detected C. trachomatis DNA in the synovium of patients with UA [9], suggesting that some of these patients may have a 'forme fruste' of ReA.
Arthritogenic bacterial DNA and RNA from Chlamydia trachomatis, Chlamydophila pneumoniae, and Yersinia pseudotuberculosis have been detected by PCR in synovial samples from patients with ReA and UA. Thus, micro-organisms, or components thereof, do reach the joint but are not always cultivable [2,[9][10][11][12]. This suggests that inflammation at the joint is caused by an immune response to bacterial antigens [9,13]. Bacterial DNA has also been detected in synovial samples from patients with other forms of arthritis, such as rheumatoid arthritis (RA) or osteoarthritis (OA) [14][15][16]. Detection of nucleic acids from other bacteria (Pseudomonas sp., Bacillus cereus, Mycobacterium tuberculosis, or Borrelia burgdorferi) in synovial fluid or synovial tissue (ST) from patients with ReA or other forms of arthritis (UA, RA, or OA) has raised the question of whether non-Chlamydia or nonenteric bacteria may enter the synovium and cause or contribute toward synovitis [14,[17][18][19]. However, the list of pathogens that trigger ReA is not definitively established.
Several studies have addressed this issue, using broad-range PCR and/or reverse transcription PCR systems to search for bacterial DNA and RNA in synovial samples from patients with various forms of arthritis, including ReA [12,14,17]. By cloning and sequencing the PCR products, they have shown that more than one micro-organism can be present in the same joint. In most studies, the PCR products were of sufficient length to determine the genus of the bacteria in the synovial samples, but were not long enough to identify the species level [12,17].
In this study we aimed to identify bacterial DNA in patients with ReA and UA using broad-range PCR, cloning and sequencing of almost the entire 16S rRNA gene. The use of this approach revealed the identity of potential bacterial causes and the presence of previously uncharacterized and uncultured bacterial pathogens in joint disease. Despite the frequent occurrence of genital and intestinal infections in Tunisia [20][21][22][23][24], no studies of ReA-related bacteria have yet been conducted in this country.

Materials and methods Patients
Twenty-eight patients with knee effusion, who had given informed consent, were included in the study after approval from our institutional review board. All patients were attending one of three rheumatology hospital departments in Tunisia. ST samples were obtained by needle biopsy from six patients with ReA (six posturethritic) and nine with UA, and from a control group of seven patients with RA and six with OA. The patients' clinical features and demographic characteristics are summarized in Table 1.
ReA was diagnosed according to European Spondyloarthropathy Study Group and Amor criteria [25,26]. All of the cases of ReA were acquired sexually, with arthritis occurring within 4 weeks of an urogenital infection (Table 1). UA was defined as a monoarthritis or oligoarthritis occurring without evidence of a predisposing infection in a patient in whom other known rheumatic diseases had been excluded.
ST samples were taken from the knee joint using the Parker-Pearson biopsy procedure [27]. Care was taken during and after obtaining patient samples to prevent cutaneous bacterial contamination. The skin surface was prepared with three successive betadine solution swabs, each for 2 minutes, and then with 70% alcohol for 2 minutes, before sampling. ST samples were immediately placed in sterile microcentrifuge tubes, which were closed and snap frozen in liquid nitrogen. Tubes were stored at -80°C until analysis.

Automated DNA extraction
A DNA extraction procedure using the MagNA Pure system (Roche Molecular Biochemicals, Meylan, France) was used for all ST samples, using a pre-extraction treatment. Before MagNA Pure extraction, 500 μl lysis buffer (200 mmol/l NaCl, 20 mmol/l Tris HCl [pH 8], 50 mmol/l EDTA, and 1% SDS) and 25 μl proteinase K (10 mg/ml; Sigma, St Louis, MO, USA) were added to approximately 10 mg of ST. The mixture was then vigorously agitated and incubated at 65°C for 30 minutes or until complete dissociation of the ST fragments. The enzymatic reaction was stopped by incubation at 95°C for 10 minutes and samples were centrifuged at 10,000 g for 5 seconds. DNA was extracted on the MagNA Pure instrument using the MagNA Pure LC DNA isolation kit-Large Volume, in accordance with the manufacturer's instructions.

Broad-range PCR amplification of 16S rRNA genes
The full-length 16S rRNA gene was amplified from extracted DNA with broad range primers (BAc08F: 5'-AGAGTTTGATC-CTGGCTCAG-3'; and Uni 1390R: 5'-GACGGGCGGTGT-GTA CAA-3'), targeting the region corresponding to nucleotides 8 to 27 and 1,390 to 1,407 of the Escherichia coli 16S rRNA gene [28,29]. DNA was amplified in 50 μl reaction mixtures, each containing 1× Ex Taq Buffer (Takara Ex taq, Otsu, Shiga, Japan), 0.2 mmol/l of each primer, 2.5 mmol/l of each DNTP, 2 mmol/l MgCl 2 and 1.25 units of Takara Ex Taq DNA polymerase (Takara Ex taq, Otsu, Shiga, Japan. T4 Gene 32 Protein (5 μg/μl; USB Corp, Cleveland, Ohio) was added to the PCR mix followed by 2.5 μl of DNA extract. PCR was performed as follows: initial denaturation at 94°C for 5 minutes, followed by 30 cycles of denaturation at 94°C for 1 minute, primer annealing at 59°C for 1 minute and extension at 72°C for 1.5 minutes. The final elongation step was extended to 15 minutes. PCR was carried out in a Gene-Amp PCR System 9700 (Applied Biosystems, Foster City, CA, USA). All extracts were tested undiluted, diluted 1:10 and 1:20, with or without T4 Gene 32 Protein, to avoid false-negative results. The T4 Gene 32 Protein was used to increase the yield of PCR products [30][31][32].
Pure DNA from either E. coli or C. trachomatis was used as a positive control for the broad-range PCR screening system. Amplification products were visualized on ethidium bromidestained 1% Seakem GTG agarose gel (Tebu-bio, Le Perray en Yvelines, France).
Precautionary measures were taken to prevent DNA contamination during DNA extraction and manipulation. These included pipeting PCR components under a laminar flow of sterile air, using only sterile equipments, dedicated pre-PCR and post-PCR rooms, and dedicated sets of pipettes, disposable gloves, laboratory coats and non-reusable waste containers. Reagents and PCR primers were aliquoted to prevent frequent handlings. DNA extraction was performed in two separated biological hoods, which were cleaned before and after each sample preparation with 5% bleach solution. Gloves were changed between each tissue sample. DNA contamination was avoided using aeroguard filter tips (TipOne; Starlab, Bagneux, France) and individually self-sealing PCR tubes (Starlab, Bagneux, France), irradiated with UV light at 254 nm for 10 minutes to inactivate extraneous DNA. Negative controls (water during the amplification step and an uninfected mouse heart tissue sample during the extraction protocol) were included every five samples for each experiment to monitor potential contamination. If amplification occurred in any of the negative controls, the PCR was repeated [33]. All samples were amplified in duplicate to allow a large number of clones to be sequenced.

Cloning, DNA sequencing and sequence analysis
The 16S rDNA amplicons were inserted into a vector using a cloning kit (pGEM-T vector; Promega, Madison, WI, USA), in accordance with the manufacturer's instructions. 16S rDNAcontaining clones were grown in Nunc microtiter plates containing 150 μl of 2 × Luria-Bertani medium supplemented with 10% glycerol and ampicillin (100 μg/ml). Insert amplifications were performed using the GE Healthcare amplification kit by the RCA (rolling circle amplification) method (GE Healthcare, formerly Amersham). Amplicons were purified and then sequenced using the commercial BigDye Terminator v3.1 kit (Applied Biosystems) on a 3730XL sequencer (Applied Biosystems). The resulting 16S rDNA clones sequences were compared to sequences in the European Molecular Biology Laboratory (EMBL) databases using BLAST (basic local align- ment search tool) and then checked for chimera using ribosomal database project II software [34].

Stastistical analysis
Data were compared by Fisher's exact test using Epi Info software, version 6.04a (Centers for Disease Control and Prevention, Atlanta, GA, USA). P < 0.05 was considered to be statistically significant.

PCR positivity by the broad-range PCR amplification system
Because PCR and extraction controls were negative, our results could be interpretated accurately. Amplification products of the 16S rRNA gene were generated from 21 of the 28 ST samples (75%) using broad-range PCR. Amplicons were detected in all samples from the six patients with ReA (100%) and nine with UA (100%). In the control group, bacterial 16S rDNA was amplified in ST samples from three of the seven patients with RA (43%) and from three of the six with OA (50% Additionally, ReA and UA samples exhibited a higher bacterial DNA load, as indicated by the signal intensity of the PCR products (Table 2). To enhance the spectrum of DNA from bacterial species detected, at least 24 individual clones from each sample were sequenced. Additional sequencing was performed if problems were encountered during the cloning of nonspecific or partial 16S rDNA products. In general, poorer DNA profiles of bacterial species were obtained from tissue samples that gave weak PCR signals (Table 2).

Bacterial 16S rDNA sequences identified in synovial tissue samples
A broad range of DNAs from bacterial species was detected in each ST sample (Table 3). Only good quality sequences, with length ≥ 1,000 nucleotides, were analyzed. Most bacterial sequences had ≥ 97% sequence similarity with cultivated or uncultured bacteria. The per cent similarity to best fit sequence from the database, the accession number and the sequence length are listed in Table 4.
DNA from a total of 68 individual bacterial species were detected in ST samples from the patients with ReA and UA, and 12 DNAs from different bacteria were identified in the control ST samples. Additionally, DNAs from 20 bacterial species were detected in both study and control samples from patients with ReA, UA, RA, or OA. Therefore, these organisms are probably common in joint diseases (Table 4). Many sequences were from commensal bacteria, in particular those normally found in the skin or the intestinal tract (Propionibacterium acnes, E. coli and other coliform bacteria). We also detected bacterial DNAs from mucosal bacterial flora such as streptococci, Actinomycetes and Neisseria, and DNAs from opportunistic pathogens such as Stenotrophomonas maltophilia, Alcaligenes faecalis, Achromobacter xylosoxidans and Acinetobacter spp. in a number of samples. We found DNAs from organisms that are commonly identified as triggering ReA, such as Shigella flexneri and Shigella sonnei [35,36], in 33.33% of ReA and UA samples, but not in control samples. S. sonnei DNA was detected in samples from one ReA and one UA patient. S. flexneri DNA was detected in samples from two patients with ReA and one with UA. DNA from Propionibacterium acnes -an arthritogenic agent involved in SAPHO (synovitis, acne, pustulosis, hyperostosis, and osteitis) syndrome, which is an oligoarthritis associated with acnes and pustilosis [37,38] -was detected in ReA and UA samples. Detection of this bacterium-derived DNA was associated with S. sonnei (patient 3) and with S. flexneri (patient 5 No genitourinary tract bacterial sequences (for example, C. trachomatis) were detected in our patient samples. This was unexpected, especially in ReA patients with a preceding urogenital infection. We also detected DNAs of several bacterial species that have previously been described in human infections but not in arthritis (

Discussion
We investigated the presence of bacterial DNA in ST samples from patients with ReA and UA, using 16S rRNA PCR, cloning and sequencing. This is, to our knowledge, the first study using the full-length 16S rRNA gene as a target for broad-spectrum PCR to detect bacterial DNA in synovial samples.
We extracted DNA from ST samples from 28 patients with arthritis. We found bacterial DNA in 21 (75%) of these patients, using stringent sterility and anti-contamination techniques. Previous studies, using PCR assays with universal 16S rDNA primers, identified lower proportions of human synovial samples containing bacterial DNA: 42% of synovial fluid and ST samples in one study [18], and 10% of ST samples in another [17]. Our high proportion of bacterial DNA in ST sam-ples from our patients may be due to the use of the primer pair (Bac08F/Uni1390R) as well as the use of the T4 Gene 32 Protein, which may increase the yield of PCR products [30][31][32].
Sequence analysis of the PCR-positive samples revealed the presence of a mixture of bacterial DNA in synovial samples from patients with ReA, UA, RA or OA. These findings are similar to those reported in previous studies [12,14,17,39]. A significant disadvantage of broad-range PCR is the tendency to yield false-positive results [33,40]. In fact, we undertook stringent precautionary measures at each step (as presented in Materials and methods; see above) to prevent contamination. In addition, the MagNAPure system used is a rapid, closed, automated and standardized method for DNA extraction, elim-       [41][42][43][44]. However, a better understanding of the contribution made by intestinal microflora to human biology is needed to elucidate the potential role played by microflora in the pathogenesis of ReA [41][42][43][44].
We detected DNAs of some bacteria that have not previously been described in human synovial samples, such as Blastococcus spp., Leucobacter lutti, Halomonas spp., Rhodococcus fascians and manganese-oxidizing bacteria (Table 4). These organisms have an environmental source (soil, plant and water). Other 16S rRNA gene sequences showing less than 97% sequence similarity with bacterial sequences from the EMBL database and affiliated with noncultivated bacteria were detected. Thus, we found DNAs from uncultured candidate division OP10 bacteria in ReA and UA samples, but not in control samples. However, the synovium is probably an interfacial zone that can be colonized by bacterial DNAs originating from the environment and the endogenous microflora [37]; indeed, only a proportion of resident commensal micro-organisms in the gut have been identified [44].
Our approach of cloning and near full-length sequencing of bacterial 16S rDNA might confirm the presence of such DNAs of bacteria not considered to be human pathogens in the synovium. Their significance remains unclear, however, and further investigations will be required to determine the pathogenic relevance of these findings.
We detected DNA from Shigella flexneri and Shigella sonnei -micro-organisms that are known to trigger ReA -in ReA and/or UA samples but not in control samples. Shigella DNA positive patients had no clinical signs of previous intestinal infection with an enteric organism. These patients may have been asymptomatic, or the preceding gastrointestinal symptoms may have been mild and overlooked by the patients [6]. It is possible that enteric organisms may move from asymptomatic primer sites of infection to the synovium in such patients [6]. Most Shigella ReA cases are caused by S. flexneri [35], but sporadic cases associated with S. sonnei and S. dysenteriae have been described [4,35]. The most recent published case of S. sonnei related ReA was attributed to sexual transmission of the pathogen [36]. In our study, we detected S. sonnei DNA in one ReA patient presenting with an urogenital  infection, which is consistent with the possibility that this species could be related to sexual transmission.
Propionibacterium acnes sequences were detected in synovial samples of one ReA patient with pustular lesions and two UA patients in whom we did not detect any other causative bacteria derived DNA. Thus, the presence of P. acnes DNA in these patients is likely to be either disease specific or due to opportunistic colonization of the inflamed joints. This bacterium is part of the normal skin flora. It was recently identified in articular samples from patients with SAPHO (synovitis, acne, pustulosis, hyperostosis, and osteitis) syndrome, which is often regarded to be a form of spondyloarthropathy, suggesting an infectious origin of this syndrome [37,38]. Our findings also suggest that this organism can access the joints.
ReA-related genitourinary bacterial species, including chlamydial sequences, were not detected in any patient samples, despite the large number of clones sequenced and analyzed for each sample. Thus, it is possible that not all bacterial DNAs present within the joint were detected. Indeed, even for arthritic patients with related urogenital infections, the detection of genital infectious agents by broad-range PCR may have been masked by the presence of other bacterial DNAs. Thus, bacteria appear to move from various anatomic sites such as gut to the joints. This is consistent with the detection of E. coli sequence in many of the synovial samples. In enteric ReA, active bowel inflammation affects the barrier function of the gut wall, allowing gut flora to access systemic sites [45]. Moreover, most of the patients were taking nonsteroidal drugs, which can impair gut permeability and mucosal competence.
We have shown that sequences from bacterial species that are known to be involved in the onset of arthritis represented a minority of the sequences detected in ST samples from patients with ReA. In addition, their presence was associated with DNAs from commensal and environmental bacterial flora. This raises the question about the role that this variety of intraarticular bacterial DNA plays in the pathogenesis of ReA and other forms of arthritides. However, detection of bacterial DNAs in the ST of patients with arthritis does not necessarily reflect the presence of complete bacterial genome, the presence of infectious bacteria or the potential of bacterial replication, or indicate whether detected DNA is related to the synovial pathology [2]. Within this context, the presence of multiple bacterial DNAs in patient joints does not substantiate a multibacterial infection that could ensue in these patients. Our detection of bacterial DNA in synovia of both control and study patient samples may indicate that a low level of 'background' bacterial DNA is usually present in synovial material and that such DNAs do not necessarily cause synovitis [17,46]. Such a variety of bacterial DNA could be due to the passive transfer of various bacterial products within phagocytic cells to the inflamed joint. Consistent with this, bacterial fragments have previously been detected -using immunohis-tochemical techniques -in macrophages from spleen of rats and humans and in synovium-derived macrophages from patients with various arthropathies [41,47,48]. Nonspecific migration of inflammatory cells containing bacterial particles into synovium could promote synovial inflammation [17]. On the other hand, bacterial DNA itself might trigger an immune response, and thus may induce synovitis. It has been shown that experimental intra-articular injection of some bacterial DNA (Escherichia coli, Staphylococcus aureus) or simply of nonmethylated CpG motifs is sufficient to trigger arthritis in mice [49][50][51]. The hypothesis is that bacterial DNA may be directly responsible for part of the synovial inflammation. This merits further investigations in humans, particularly to assess whether the quantity of bacterial DNA required to induce arthritis is comparable with the 'inoculum' observed in vitro in ReA and other arthritides.

Conclusion
Broad-range PCR, sequencing and cloning are essential techniques for the characterization of the microbial environment in joints. This is the first study to use a broad-range 1,400 base pair 16S rDNA PCR coupled to automated DNA extraction to identify bacterial nucleic acid present in the joints of patients with ReA and other forms of arthritis. Our study provides a potential overall picture of the detailed presence of bacterial DNA in arthritic joint. Cloning procedures allowed the identification of known ReA-triggering organisms, unknown pathogens, and potentially novel bacterial species not previously associated with ReA and other forms of arthritis