Development of a Real-Time PCR for Identification of Brachyspira Species in Human Colonic Biopsies

Background Brachyspira species are fastidious anaerobic microorganisms, that infect the colon of various animals. The genus contains both important pathogens of livestock as well as commensals. Two species are known to infect humans: B. aalborgi and B. pilosicoli. There is some evidence suggesting that the veterinary pathogenic B. pilosicoli is a potential zoonotic agent, however, since diagnosis in humans is based on histopathology of colon biopsies, species identification is not routinely performed in human materials. Methods The study population comprised 57 patients with microscopic evidence of Brachyspira infection and 26 patients with no histopathological evidence of Brachyspira infection. Concomitant faecal samples were available from three infected patients. Based on publically available 16S rDNA gene sequences of all Brachyspira species, species-specific primer sets were designed. DNA was extracted and tested by real-time PCR and 16S rDNA was sequenced. Results Sensitivity and specificity for identification of Brachyspira species in colon biopsies was 100% and 87.7% respectively. Sequencing revealed B. pilosicoli in 15.4% of patients, B. aalborgi in 76.9% and a third species, tentatively named “Brachyspira hominis”, in 26.2%. Ten patients (12.3%) had a double and two (3.1%) a triple infection. The presence of Brachyspira pilosicoli was significantly associated with inflammatory changes in the colon-biopsy (p = 0.028). Conclusions This newly designed PCR allows for sub-differentiation of Brachyspira species in patient material and thus allows large-scaled surveillance studies to elucidate the pathogenicity of human Brachyspira infections. One-third of affected patients appeared to be infected with a novel species.


Introduction
Brachyspiraceae are fastidious anaerobic spirochaetes and are the causative agents of intestinal spirochaetosis, a condition that is characterized by the attachment of Brachyspira species to the colonic mucosa.
In animals, Brachyspira pilosicoli, B. hyodysenteriae, B. intermedia and B. alvinipulli can cause serious disease [1][2][3], whereas B. innocens and B. murdochii are considered harmless commensals of the gut-flora. [4] Brachyspira pilosicoli infects a wide range of mammals, including humans [5,6] , and is the undisputed etiological agent of porcine intestinal spirochaetosis [3,7], and a well known pathogen causing diarrhoea and dysentery in pigs and chickens. [8][9][10] In humans thus far only two infecting species have been reported in detail: Brachyspira aalborgi (which is found exclusively in humans and higher primates) [11,12] and B. pilosicoli (suggesting putative zoonotic potential of B. pilosicoli). [8,13] Additionally, there is limited molecular evidence of a third Brachyspira species that awaits formal confirmation. [14][15][16][17] Jensen et al. [15] used fluorescent in situ hybridization to identify an uncharacterised Brachyspira species and proposed the name ''Brachyspira christiani'', but since they did not sequence 16S rDNA, it can not be ascertained that this species is the same as the species describe by Pettersson et al. [16] We therefore propose to refer to the species described in this paper as ''Brachyspira hominis'', as it has currently only been described in humans.
Symptoms commonly associated with human intestinal spirochaetosis are chronic diarrhoea, abdominal pain, weight loss, faecal mucus and blood-stained faeces but the clinical relevance of these findings has been debated. [18][19][20] Invasive spirochaetaemia has been reported in critically ill and/or immuno-compromised patients. [21,22].
The pathogenic potential of these bacteria in humans has never been thoroughly investigated. Due to the strict anaerobic growth requirements and long incubation times it is difficult to routinely culture Brachyspira species. [12,23] Therefore, routine diagnosis of human intestinal spirochaetosis is currently only based on histopathology of colonic biopsies (figure S1) and the absence of distinct morphological hallmarks prohibit the discrimination between pathogenic and non-pathogenic Brachyspira species. In addition, the invasive nature of colonoscopy limits epidemiological as well as non-invasive clinical follow-up studies. As a result not much is known on the species distribution of Brachyspiraceae in human infections. Based on histopathological examination of colon biopsies obtained for various reasons the estimated prevalence in the general Western population varies from 1.1% to 6.9%, with higher incidences among immunocompromised patients. [24][25][26] Some studies have been performed using speciesspecific primers, investigating the epidemiology of B. aalborgi and B. pilosicoli in humans. Based on these, B. aalborgi causes human infection more frequently (80%-88.6%) than B. pilosicoli (5.7%-14.3%), however, no data exists on infections with ''B. hominis''. [27][28][29].
Several PCRs for detecting Brachyspira in faecal or cadaver specimens of animals have been published. [30] These are neither validated for human diagnostic use, nor allow for immediate species determination. To our knowledge, only one PCR has been described for use on human faeces. [31] However, due to the size of the amplified fragment, this PCR is neither suited for use on formalin-fixed paraffin-embedded biopsy samples nor does it allow for real-time species identification. In addition, these PCRs have been species specific, thus only amplifying 16S rDNA from one species and do not allow for the identification of multi-infections with hitherto unknown Brachyspira species.

Aims of this Study
In this study we will describe the design and validation of a realtime PCR allowing species differentiation of Brachyspira species in formalin fixed paraffin tissue specimens and faeces. In addition we investigate species distribution and symptoms in a cohort of patients with histologically documented Brachyspira infection.

Ethics Statement
Patients visiting a Dutch hospital are actively informed of the 'opt-out' system regarding research on archival patient material. Dutch law requires all studies using such materials to obtain an official approval by the local ethics committee. This study was approved by the Medical Ethics Testing Committee of the University Medical Centre Utrecht, The Netherlands, under protocol number 11-402/C. Only material from patients who did not opt-out has been included in this study.

Sample Selection
Formalin-fixed paraffin-embedded (FFPE) colon tissue taken between 2001 and 2011 was retrospectively identified in four Dutch hospitals (two regional hospitals (Tergooiziekenhuizen, Blaricum/Hilversum and St. Antonius Hospital, Nieuwegein) and two large academic centres (the University Medical Centre,  Utrecht and Vrije Universiteit Medical Centre, Amsterdam). Samples were included if the pathology report reported the presence of spirochaetosis and if this was confirmed by immunohistochemistry (Borrelia burgdorferi (Lyme) IgG antiserum, ILP 0301, ImmunoLogic, Duiven, The Netherlands). This staining is based on the immunological cross-reactivity of a universal spirochaete antigen and is used routinely in most Dutch laboratories to confirm the diagnosis of spirochaetes in histological samples. As control samples, spirochaete negative colon biopsies, taken between January 2009 and December 2010, were selected if tissue quality was macroscopically good and no spirochaetes were detected by immunohistochemistry.
All colon-biopsy materials were revised by two experienced gastrointestinal pathologists (MEIS and HVS), who were unaware of the initial diagnosis and PCR-results. The presence or absence of inflammatory cells in spirochaete-positive biopsy samples was scored by one pathologist (HVS). Where possible, relevant clinical data were retrieved from clinical files of patients with Brachyspira infection (i.e., sex, age at time of biopsy, indication for biopsy, HIV-status, and presence or absence of rectal blood loss, diarrhoea, abdominal pain, weight loss or faecal mucus).
From three patients, faecal samples were obtained. DNA from these samples was isolated using the same protocol as for the FFPE samples, and further analysis was identical to the colon biopsies.

Processing of FFPE Biopsy Samples
Of each FFPE-sample, three 4 mm sections were cut for pathological evaluation, and, subsequently, five 20 mm sections were cut and placed in a 1,5 ml Eppendorf tube (Eppendorf, Hamburg, Germany), and three additional control sections were cut (4 mm). The knife was then discarded and the microtome thoroughly cleaned with 96% ethanol to prevent the next FFPEsample to be contaminated with DNA from the previous sample. The five 20 mm sections were frozen at 280uC until DNA isolation. The six 4 mm sections were fixed on glass slides and archived at room temperature.

DNA-isolation
DNA-isolation from FFPE samples was performed by standard procedures. These consisted of the removal of the paraffin material through several wash steps followed by automated DNA extraction. The first wash step consisted of adding 1 ml of xylene and fifteen minutes incubation in a shaking heat block (1,400 rpm, 22uC). The samples were then centrifuged at 13,000 g for five minutes and subsequently the xylene was removed. This xylene wash-step was performed twice. Then 1 ml 96% ethanol was added to the pellet and samples were vortexed for 5 seconds. Subsequently, samples were centrifuged at 13.000 g for five minutes after which the ethanol was discarded. The ethanol wash-step was performed twice. The final wash-step consisted of adding 1 ml acetone, vortexing for 5 seconds, centrifuging at 13.000 g for 5 minutes and carefully removing the acetone. The pellet was then allowed to dry for 30 seconds at room temperature. Subsequently, 360 ml digestion buffer (10 mM hydroxymethylaminomethane, 1 mM ethyleendiamine tetracetic acid, 0.001% sodium dodecyl sulphate and 0.5% Triton X 100) and 20 ml protein K (600 U/ml, Roche Diagnostics, Almere, The Netherlands) was added to the pellet. Finally 40 ml Phocine Herpes Virus (PhHV) DNA was added as an internal control to monitor the DNA isolation and PCR amplification processes. [32] The resultant suspension was placed in a shaking heat block (750 rpm, 70uC) for 30 minutes to allow digestion of tissue and subsequently incubated at 95uC for 15 minutes and 750 rpm to inactivate the protein K. The samples were then stored at 280uC until DNA isolation. DNA was isolated using the MagNA Pure 96 DNA and Viral NA Small Volume Kit on a Roche MagnaPure96 (Roche Diagnostics, Almere, The Netherlands) with the Viral NA Universal SV extraction protocol according to the manufacturer's instructions.
For the isolation of DNA from faecal samples a suspension of 25 mg faeces in 200 ml lysis buffer (cobasH PCR Female Swab Sample Kit, Roche Diagnostics, Mannheim, Germany) was prepared and spiked with 40 ml PhHV DNA, mixed well and placed at 280uC for more than 2 hours. The frozen solution was then incubated at 95uC for 30 minutes and DNA was extracted using the MagNA Pure 96 DNA and Viral NA Small Volume Kit on a Roche MagnaPure96 (Roche Diagnostics, Almere, The Netherlands), with the Viral NA Universal SV extraction protocol according to the manufacturers instructions.  [33] and GenBank (Release 181.0 (14/12/2010)) and aligned using Seaview (version 4.2.8). [34,35] Four Brachyspira-specific primer-sets were designed based on this alignment (table 1) and analysed in silico for undesirable targets with the Probe Match function of the Ribosomal Database Project and the Primer-BLAST-algorithm. [36] This primer-set was designed to amplify an 82 base-pair region that is specific for both B. pilosicoli, B. aalborgi, and the putative ''B. hominis'' with the conditions that 1) the primer binding sites are conserved in all Brachyspira species to ensure that all Brachyspira species will be amplified, and 2) the primer binding sites differ sufficiently in sequence from all other known bacterial 16S rDNA sequences so that only 16S from Brachyspira species will be amplified, and 3) the region between these primers differs in length and/or base-pair composition between the three human Brachyspira species thus generating sufficient differences in melting-temperatures of the PCR fragments, allowing for a melting curve based species identification. This was verified with Clone Manager 9 Professional Edition (version 9.0), which uses the expression by Freier et al. [37] Subsequently, this theoretical difference was confirmed using the available human reference strains (B. pilosicoli (WesB) and B. aalborgi (513A T ). Since ''B. hominis'' has only been described molecularly no (type or reference) strains where available, and therefore it was not possible to test the correctness of the ''B. hominis'' melting temperature prediction. Three sets were designed for the generation of Brachyspira specific 16S rDNA fragments thus allowing to obtain the sequences of the first 517bp of 16S rDNA (StepA, StepB and StepC) of the infecting Brachyspira species from the DNA isolated from the biopsies.

Primer-design
Based on results obtained in this study, six primer sets were designed to allow for the generation of Brachyspira species-specific 16S rDNA fragments in double-infected samples for sequencing

Sequencing
To allow for the sequencing of the Brachyspira species three primer-sets (Step A, B and C; table 1, figure 1) were designed, that, when assembled, cover the first 517bp of the 16S rDNA. In case of a double infection, species-specific primers (table 1) were used to generate separate PCR products for the individual Brachyspira species predicted to be present in these samples. Sequencing of the PCR products was performed using the BigDye Terminator v1.1 technique on a 3730 DNA Analyser (Applied Biosystems, Carlsbad, California, United States of America) according to the manufactures instructions. Sequences were manually assembled

Construction of Phylogenetic Tree
The assembled sequences and the type strains from all Brachyspira species obtained from the Ribosomal Database were aligned with the aid of Seaview (version 4.2.8) [34,35,39] and a phylogenetic tree was constructed from these alignments with SplitsTree (version 4.11.3, built February 26, 2010) [40] using Borrelia burgdorferi (ATCC 35210) as an out-species.

PCR Validation
The

Sequencing, Melting-curve Analysis and Phylogenetic Analysis
The Bspp primer-set produced a product in all samples from histopathologically positive patients and no product was produced in negative controls. Interpretation of the melting-curves is shown in table 2, compared to sequence results. With the exception of two samples (patient 6-II and patients 57), all PCR-positive samples were successfully amplified using primer-sets StepA-C, thus allowing the assembly of the first 517bp of the 16S rDNA. Of patients 6-II and 57, the central part (StepB) of the sequence could not be amplified despite repeated attempts. However, the first (59-)part could be obtained (StepA) and this contains the most speciesspecific information, allowing for proper species identification. Based on the sequence results, B. aalborgi was identified in 50 (76.9%), B. pilosicoli in 10 (15.4%) and ''B. hominis'' in 17 (26.2%) samples. There were eight (12.3%) double infections and two (3.1%) triple infections (table 2). The obtained sequences were submitted to GenBank (accession numbers JX446497 through JX446571, table S1). Using the complete 517bp 16S rDNA sequences a phylogenetic tree (figure 3) was constructed, with the addition of the sequences described by Pettersson and colleagues (named Hca, Hcc, Tva and Tvb followed by a number). [16] In this tree, two lineages consisting of novel Brachyspira sequences can be indicated (as per Pettersson et al.: cluster 1 (B. aalborgi), cluster 2 (''B. hominis'') and cluster 3 (Brachyspira species). Patient 44 differs from the type strain of B. aalborgi by 14 base-pairs, and is closest related to Tvb03, with a difference of nine base-pairs.
When comparing sequence results to melting-curves, melting curves correctly identified the species present in 57 of 65 FFPE samples. In eight samples a double infection and in two samples a triple infection was predicted and in all cases confirmed by sequence analysis. No discrepancies were observed between the duplicate PCRs of the samples. Misclassification occurred in eight samples; six predicted ''B. hominis'' appeared B. aalborgi on sequence analysis and two predicted B. pilosicoli's appeared ''B. hominis'' (Table 2).

Correlation between the Infecting Species and Clinical Symptoms
Clinical information was available from 49 patients with Brachyspira infection (table 3). No significant associations were found except that inflammatory changes were more frequently observed in the presence of B. pilosicoli (p = 0.028; Chi-squared test).

Discussion
We have designed a real-time PCR for species-specific identification of Brachyspira species in formalin-fixed paraffinembedded material and faeces, which now allows large-scale epidemiological studies of this human disease. The first analysis of 57 infected patients revealed a high prevalence of a novel speciestentatively called -''B. hominis'' and a large proportion of double infections.
We designed a Brachyspira species-specific PCR targeting the 16S rDNA that can identify all Brachyspira species that have been associated with infection in humans. Pettersson and colleagues identified three groups of spirochaetes that were closely related to, but based on 16S rDNA, phylogenetically different from, Brachyspira aalborgi. [16] These findings have been confirmed, but no epidemiological data exist. [14,15,17] Next to the 16S rDNA, the NADH oxidase gene (NOX gene) is frequently used to identify Brachyspira species. [41,42]  Several authors have remarked that the 16S rDNA genes of the various Brachyspira species are highly similar, and that species differentiation based solely on 16S is difficult. [17,43,44] Therefore, finding a 16S sequence that differs from known species is a strong indicator that this in fact is a novel Brachyspira species. It is for this reason that the NOX gene is frequently used to identify the species of Brachyspiraceae. The described primers by Mikosza et al. target B. aalborgi and ''B. hominis'' sequences only, not amplifying any other Brachyspira species, therefore limiting direct species differentiation in human material, since B. pilosicoli also occurs in humans. Further studies are needed to better classify these novel sequences.
In 65 samples of 57 patients with histologically documented Brachyspira-infection we identified 17 (26.2%) samples with this novel 16S rDNA sequence of ''B. hominis'', suggesting that more than the two hitherto described Brachyspira species are involved in human intestinal spirochaetosis. Moreover, these novel sequences were frequently (10/17; 58.8%) present in double-or even tripleinfections, which may explain why they have not frequently been identified before. Our finding that this species was present as a mono-infection in seven patients and as a co-infection with other Brachyspira species in ten patients strongly suggest that this novel Brachyspira species is capable of infecting humans. However, more studies are needed to confirm these preliminary findings.
Ribosomal DNA sequencing is considered the gold standard in species determination. Our Bspp PCR provides rapid information with 100% accuracy for the presence (or absence) of Brachyspira species, and with 87.7% accuracy (95% confidence interval 0.76-0.94) for species identification. However, there were eight (out of 65; 12.3%) discrepancies between the predicted species based on the melting-curve and the sequence result. The melting-curves rely on a ,1uC difference between species (figure 2; ,79uC for B. pilosicoli, ,80uC for ''B. hominis'' and ,81uC for B. aalborgi). There are several known issues with SYBR-green. The T m is influenced by a variety of factors. [45] Impurities in the extracted DNA, and/ or higher concentrations of DNA might result in a minute shift of the melting temperature: there were never any discrepancies between the individual duplicates of samples. It is further supported by the observation that the melting-curve never shifted to the left, which would indicate a lower melting-temperature, but always to the right, indicating a higher melting-temperature (table 2). This is a known issue of SYBR-green based meltingcurves. [45].
We acknowledge that the only true way of definitively calling a new species is to culture the bacterium. This is indeed a limitation that we are currently unable to solve, as we could not culture Brachyspira species from the formalin-fixed samples that were retrospectively identified in this study.
Based on the three Brachyspira positive faecal samples that were present in this study that could be correlated to positive biopsy samples, we conclude that this PCR allows for the testing of faeces. As this observation is based on concomitant material (biopsy and faeces) of three patients only, further studies are needed.
Although the analysis of 49 patients with Brachyspira infections suggested no correlation between species and physical complaints, the high prevalence of double infections, the small sample size of a the biased population (only patients of whom biopsies were taken for analyses of gastro-intestinal complaints) precludes strong conclusions on the clinical spectrum of disease caused by Brachyspira species. However, in pigs and poultry, B. pilosicoli is a known pathogen and the fact that the presence of B. pilosicoli was significantly associated with inflammatory changes in the colonbiopsy might suggest an immunologic reaction of the host against B. pilosicoli. Since only 49 patients had complete clinical records, no definitive pathogenic potential in humans can be attributed to B. pilosicoli. Further research into this finding is necessary.
In conclusion, we designed a real-time PCR for the immediate differentiation of Brachyspira species in human sample material, allowing for epidemiological studies in formalin-fixed paraffinembedded material and faeces as well as routine, non-invasive clinical follow-up. Using this PCR, we investigated the epidemiology of the different Brachyspira species, and found a relatively high incidence of the novel ''B. hominis'' and a large proportion of double infections. Figure S1 Histopathological image of human intestinal spirochaetosis. The spirochaetes are present as a 'false brush border' on the mucosal surface of the entire colon, as indicated by the arrow (haematoxylin and eosin stain, original magnification 630 times, bar equals 20 mm).