INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Whipple's disease is a systemic bacterial disease responsible for low-grade fever, weight loss, diarrhea, lymphadenopathy, polyarthritis, and occasionally cardiac involvement (37, 38). Although its bacterial nature was demonstrated by electron microscopy in 1961 (40), only very recently has the Whipple's disease bacillus been isolated and propagated in the laboratory (27).

Molecular data from the Whipple's disease bacillus have been limited to the amplification and sequencing of a few genes directly from infected human tissues (39) or environmental specimens (17), and only ribosomal sequences are available from the sole Whipple's disease bacillus isolate (14, 27). Based on partial 16S ribosomal DNA (rDNA) sequence analysis, the phylogenetic position of Tropheryma whippelii has been found to be within the actinomycetes (28, 39). Subsequent determination of a nearly complete 16S rDNA sequence and the 16S-23S rRNA intergenic spacer allowed a reassessment of this position as lying between the clade made up of actinomycetes possessing group B peptidoglycan and the family Cellulomonadaceae (16). These taxonomic relationships were recently confirmed by analyses of sequences derived from the actinobacterial insertion in domain III of the 23S rDNA (8) and hsp65 (22). Taxon-specific 16S rDNA primers have since been used to detect the bacterium in patients (28, 39) and sewage effluent (17), while sequence analysis of the 16S-23S rDNA spacer was used for the differentiation of strains into three different groups on the basis of their genotypes (9). Two additional genotypes were also subsequently described (18).

This consensus PCR approach has been limited to the study of a few conserved genes, and, with the exception of the 16S-23S spacer region, only partial sequences have been analyzed. Furthermore, a recent study has indicated that two different genotypes could be detected in the same clinical specimen (18), thereby introducing the possibility of cross-contamination and thus the determination of erroneous sequences when infected tissues are examined directly. The availability of the first T. whippelii isolate (27) has allowed us to apply the genome walking strategy (32) to the accurate determination of gene sequences in this emerging bacterial species. The rpoB gene was chosen as a suitable initial target for the following reasons. rpoB encodes the -subunit of RNA polymerase, an enzymatic complex conserved among Bacteria and Archaea (13). Comparison of rpoB sequences has previously been used for phylogenetic inference among Archaea and some Bacteria (21, 31). Partial rpoB sequence analysis has been shown to be a powerful tool for the accurate identification of enteric bacterial species (20), Mycobacterium spp. (12), spirochetes (30) including Borrelia burgdorferi (15), Bartonella spp. (29), Coxiella burnetii (21), and Rickettsia spp. (4). rpoB has been targeted in the first commercialized DNA chip-based identification scheme for use in a clinical laboratory (6). Additionally, the investigation of alternative molecular targets will result in further genetic characterization of the T. whippelii strains, enhancing our understanding of the epidemiology and pathogenicity of the disease. Finally, rpoB mutations have been associated with rifampin resistance in various bacterial species including Escherichia coli (11), Mycobacterium spp. (10, 35), Neisseria meningitidis (2), Staphylococcus aureus (1), Streptococcus pneumoniae (5, 24), and Rickettsia spp. (4). Rifampin has been advocated as a first-line drug for the treatment of Whipple's disease (33, 37), and thus it may be of value to develop an rpoB sequence-based detection method for rifampin resistance in clinical specimens.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains. T. whippelii Twist strain, a strain with a type 2A genotype (14), was cocultivated with HEL cells in 150-cm² flasks as previously described (27) until its 20th passage. The absence of mycoplasma contamination was checked by using the mycoplasma detection kit (Boehringer GmbH, Mannheim, Germany). Cellular infection was monitored by microscopic examination of Gimenez-stained cells scraped from the flasks. When a heavy infection was seen, the supernatants of 10 flasks were removed and mixed with the infected cells released from each flask surface by brief treatment with 0.5% trypsin (Gibco-BRL, Cergy-Pontoise, France). After centrifugation (5,000 × g for 15 min), the pellet was resuspended in 20 ml of K36 buffer (2% KH₂PO₄, 6% K₂HPO₄, 7.4% KCl; 0.9% NaCl) and incubated for 45 min at 30°C in the presence of trypsin (0.5%, final concentration); cells were further disrupted by vortexing and passages through an 18-gauge needle. Disrupted cells were centrifuged on 25% sucrose at 5,000 × g for 30 min, and the pellet was washed in K36 and further incubated for 45 min at 30°C in the presence of trypsin (10 mg/ml, final concentration). After being washed in K36 buffer, the pellet was centrifuged on Gastrografin (Schering, Lys-Lez-Lannoy, France) and then washed twice in K36 buffer and stored at 70°C until used.

rpoB amplification and sequence. DNA was extracted by using the DNA extraction kit and the Fast-prep DNA device with CLS-TC lysis buffer as described by the supplier (Bio 101 Inc., La Jolla, Calif.) with the exception that silica-adsorbed DNA was washed twice. rpoB was amplified by combining a consensus PCR approach and a genome walking approach. Consensus PCR primers (Table 1) were designed after alignment of bacterial rpoB and RpoB sequences in GenBank. Primer pair D4U-R7U was designed after alignment of all bacterial RpoB sequences, primer pairs Wh4F-Bal7R and myco11F-Wh1200R were designed after alignment of Mycobacterium smegmatis, Mycobacterium leprae, and Mycobacterium tuberculosis rpoB, and primer pair Bal1100F-Wh4R was designed on the basis of alignment of the Bacillus licheniformis, Salmonella enterica serovar Typhimurium, and M. tuberculosis rpoB genes. PCRs were performed using a Perkin-Elmer 9600 thermocycler under the following conditions. Following a first denaturation step (95°C for 2 min), a three-step cycle of 94°C for 30 s, 50°C for 30 s, and 72°C for 1 min was repeated 35 times. The final stage of the PCR program was a single 3-min extension step at 72°C. The PCR mixture incorporated 10 ng of DNA, 10 pmol of each primer, 0.2 mmol of each deoxynucleoside triphosphate, 5 µl of Taq buffer, and 2 U of Taq polymerase (Gibco-BRL) in a final volume of 50 µl adjusted with sterile distilled water. Each PCR included distilled sterile water as the negative control and DNA extracted from noninoculated HEL cells as a control for specificity. Completion of the rpoB sequence was achieved by genome walking using the GenomeWalker kit (Clontech Laboratories, Palo Alto, Calif.), a procedure which allows the creation of uncloned libraries from genomic DNA extracted following Fast-prep procedures (Bio 101 Inc.). Briefly, DNA was digested with four different restriction enzymes to obtain blunt ends and, following purification of the DNA fragments, each DNA fragment was ligated to a GenomeWalker adapter (32). PCRs were then performed using an adapter primer supplied by the manufacturer and an rpoB gene-specific primer. Specific primers selected on the basis of the ongoing sequence as being as close as possible to the known 5' extremity of the gene permitted upstream walking, whereas a primer selected close to the 3' end permitted downstream walking (Table 1). This amplification step was performed using elongase purchased from Boehringer GmbH. Following a first denaturation step (95°C for 1 min), a three-step cycle of 94°C for 30 s, 60°C for 30 s, and 68°C for 2 min was repeated 35 times. The final stage of the PCR program was a single 3-min extension step at 68°C. The success of each PCR was assessed by UV illumination of ethidium bromide-stained 1% agarose gels after electrophoresis. The resulting amplicons were purified (QIAquick spin PCR purification kit; Qiagen S.A., Courtaboeuf, France) and then sequenced using the reagents of the ABI Prism dRhodamine dye terminator cycle sequencing ready reaction kit (Perkin-Elmer Applied Biosystems, Foster City, Calif.) according to the manufacturer's instructions and using the following thermal program: 25 cycles consisting of denaturation at 95°C for 20 s, primer annealing at 50°C for 10 s, and extension at 60°C for 2 min. Products of sequencing reactions were resolved by electrophoresis in a 0.2-mm-thick 6% polyacrylamide denaturing gel and recorded using an ABI Prism 377 DNA sequencer (Perkin-Elmer Applied Biosystems) in accordance with the standard protocol of the supplier. The results obtained were processed into sequence data by sequence analysis software (Applied Biosystems), and then partial sequences were combined into a single consensus sequence.

rpoB sequence data analysis. Bacterial rpoB sequences of non-Whipple's disease bacilli were obtained from the GenBank database under the following GenBank accession numbers: Amycolatopsis mediterranei, AF242549; Aquifex pyrophilus, X75046; Bacillus subtilis, L43593; B. licheniformis, AF172323; Bartonella henselae, M73229; Bartonella quintana, AF165994; Borrelia burgdorferi, AE001144; Campylobacter jejuni, AF068778; Chlamydia muridarum, AE002327; Chlamydia pneumoniae, AE001593; Chlamydia trachomatis, AE001304; Coxiella burnetii, U86688; E. coli, U76222; Halobacterium halobium, X57144; Helicobacter pylori, M88157; Leptospira biflexa, AF150880; M. leprae, Z14314; M. smegmatis, U24494; M. tuberculosis, L27989; Mycoplasma gallisepticum, L38402; Mycoplasma genitalium, U39715; Mycoplasma pneumoniae, AE000030; N. meningitidis, Z54353; Porphyromonas cangingivalis, Y16470; Rickettsia prowazekii, AF034531; Rickettsia typhi, P77941; S. enterica serovar Typhimurium, X04642; Spiroplasma citri, U25815; Staphylococcus aureus, U970062; Synechocystis sp., D90905; Thermotoga maritima, X72695; Treponema pallidum, AE001205; Ureaplasma urealyticum, AE002118. Pairwise sequence comparisons were determined using the GCG program (Infobiogen). The sequences were aligned by using multisequence alignment program CLUSTALW, version 1.8 (36), in the DNA Data Bank of Japan (Mishima, Japan [http://www.ddbj.nig.ac.jp]). The distance matrices for the aligned sequences with all gaps ignored were calculated using the Kimura two-parameter method, and the neighbor-joining method was used for constructing a phylogenetic tree. Evaluation of individual node strength used the same program with 100 samples. Tree figures were generated using the Tree View program, version 1.61. (25).

Molecular detection and identification of T. whippelii in clinical samples. Ten jejunal biopsies were blindly subjected to rpoB-based detection of T. whippelii. These samples were composed of three biopsies previously demonstrated to contain T. whippelii DNA by 16S rDNA-based amplification and sequencing and seven negative controls. The three positive samples comprised two type 1A strains and one type 2A strain. Total DNA was extracted from each biopsy specimen using the Tissue Qiagen kit (Qiagen), and 5 µl of extracted DNA was incorporated into a PCR mixture including 10 pmol of each primer (TwrpoB.R, 3'-GCA CCG CAA CCT CGG AGA AA-5' [positions 713 to 745 in the T. whippelii rpoB coding sequence], and TwrpoB.F, 3'-TTG AGC GCA CGC CGG AAA AA-5' [positions 1181 to 1200 in the T. whippelii rpoB coding sequence]; designed to specifically amplify a 650-bp fragment of T. whippelii rpoB after alignment of rpoB sequences of T. whippelii, M. tuberculosis, M. smegmatis, and A. mediterranei), 0.2 mmol of each deoxynucleoside triphosphate, 5 µl of Taq buffer, and 2 U of Taq polymerase (Gibco-BRL) in a final volume of 50 µl adjusted with sterile distilled water. PCR conditions were as follows. Following a first denaturation step (95°C for 2 min), a three-step cycle of 94°C for 30 s, 58°C for 30 s, and 72°C for 1 min was repeated 35 times. The final stage of the PCR program was a single 3-min extension step at 72°C. Each PCR included distilled sterile water as the negative control.

Nucleotide sequence accession number. The Whipple's disease bacillus strain Twist rpoB sequence was deposited in the GenBank database under accession no. AF243072.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

T. whippelii rpoB sequence. The consensus PCR approach allowed amplification of a 2,750-bp sequence in four overlapping fragments (Fig. 1). The first fragment (F1; 1,400 bp) was amplified using the D4U-R7U primer pair designed from a comparison of proteins encoded by all bacterial rpoB sequences available in GenBank. Sequencing was achieved only with the D4U primer. From this partial sequence, we designed T. whippelii-specific primers Wh4F and Wh4R. Since the F1 fragment showed significant similarities with rpoB genes from B. licheniformis (Bal) and M. leprae and M. tuberculosis (myco), we designed primers specific to these genes, anticipating that some would be able to hybridize with the rpoB gene from T. whippelii. Indeed, the Bal7R primer, together with Wh4F, allowed the amplification and sequencing of the F2 fragment (700 bp), while Bal1100F together with Wh4R allowed the amplification and sequencing of the F3 fragment (800 bp). From this F3 fragment, we designed the Wh1200R and Wh3R (T. whippelii-specific) primers. Again, we were able to amplify and sequence the larger F4 fragment (1,700 bp) using either the myco11F or myco12F primer in conjunction with Wh1200R. No amplification was obtained from negative controls. The two extremities of the gene were sequenced using genome walking (Fig. 1). For this approach, the TwF7-1 and TwR2-1 (T. whippelii-specific) primers were designed, allowing the amplification and sequencing of the F5 (500-bp) and F6 (500-bp) fragments, respectively. From the F6 fragment, we designed the TwR2-4 (T. whippelii-specific) primer, which allowed the amplification and sequencing of the F8 (1,100-bp) fragment. From the F5 fragment, we designed the TwF7-5 (T. whippelii-specific) primer, which allowed the amplification and sequencing of the F7 (1,500-bp) fragment. Overall, we amplified and sequenced a 5,804-bp fragment comprising 983 bp upstream of the rpoB gene, the entire rpoB gene, and 1,164 bp downstream of the rpoB gene. The upstream fragment contained 1,118 bp of the 3' end of the T. whippelii rpoC gene. The putative T. whippelii rpoB open reading frame (ORF) was found to comprise 3,657 bp between the ATG start codon and the TAG stop codon. The start codon was preceded by purine-rich sequence GTCCTG, similar to the typical consensus ribosome binding site (CTCCTC). The calculated guanosine and cytosine content was 50.4%. The rpoB ORF was putatively translated into a protein of 1,218 amino acids with a calculated molecular mass of 138 kDa and a theoretical isoelectric point of 6. 

View larger version (52K): [in this window] [in a new window]   FIG. 1.   (A) Primers, amplification, and genome walking systems used to analyze T. whippelii rpoB. (B) T. whippelii DNA was extracted either from purified bacteria (lanes 1 to 4 and 10 to 13) or from cultures of bacteria grown on human cells (lanes 5 to 8 and 14 to 17). DNA libraries were obtained after enzymatic restriction and ligation to a universal adapter; enzymatic restriction was with EcoRI (lanes 1, 5, 10, and 14), DraI (lanes 2, 6, 11, and 15), PvuII (lanes 3, 7, 12, and 16), and SspI (lanes 4, 8, 13, and 17). Amplification was obtained using a primer hybridizing to the universal adapter; the second primer is rpoB specific: TWR2-4 is specific to the 3' rpoB region (lanes 10 to 17) and TWF7-5 is specific to the 5' rpoB region (lanes 1 to 8). Lane 9, molecular weight marker VI (Boehringer GmbH).

Comparison of the Whipple's disease bacillus rpoB sequence and phylogeny. When compared with RpoB sequences available for other Bacteria, that of the Whipple's disease bacillus was found to be most similar to those of other Proteobacteria, ranging from 57.19 (M. pneumoniae) to 74.63% (M. tuberculosis). Lower similarities to other groups were found. A protein multiple alignment was derived as a basis for inferring protein-based phylogenies. Neighbor-joining methods, whatever the distance algorithm used, and parsimony analysis resulted in reconstructions similar to nucleic acid-based phylogenies with significant bootstrap values at all nodes. These reconstructions were in agreement with those inferred from 16S rDNA analyses and supported a phylogenetic position for T. whippelii on a branch derived from the node which also supports A. mediterranei, M. leprae, and M. tuberculosis (Fig. 2). This phylogenetic position was supported by bootstrap values of 100%.

View larger version (30K): [in this window] [in a new window]   FIG. 2.   RpoB-based reconstructions of phylogenetic relationships of T. whippelii strain Twist. The tree was constructed by a neighbor-joining method. Scale bar, 1 inferred amino acid substitution per 100 residues. Numbers at branching points indicate bootstrap values >90%.

Molecular detection and identification of T. whippelii. An amplicon of the expected 507-bp size was obtained in three of three positive samples and zero of seven negative specimens (Fig. 3). The sequence derived from each amplicon exhibited 100% similarity with that determined for the Twist strain of T. whippelii regardless of the genotype of the strain under investigation.

View larger version (40K): [in this window] [in a new window]   FIG. 3.   Detection of the T. whippelii rpoB gene directly from human jejunal biopsy specimens using primers TwrpoB.F and TwrpoB.R, generating a 650-bp product. Lanes 2 through 4, specimens from proven T. whippelii-positive patients; lanes 6 through 13, specimens from T. whippelii-negative patients; lane 14, negative control (highly pure water); lanes 1, 5, and 15, molecular weight marker VI (Boehringer GmbH).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The entire rpoB sequence for the first Whipple's disease bacillus was easily determined using the genome walker approach, a new procedure allowing determination of unknown genomic sequences adjacent to a known one without molecular cloning (32). This approach has been previously applied in our laboratory to the determination of the rpoB gene in Bartonella spp. (29) and is radically different from the consensus PCR approach previously used to explore the Whipple's disease bacillus genome. The genome walking approach is more rapid than the consensus PCR approach, allows start and stop codons to be obtained, and, unlike the consensus PCR approach, can be applied even to moderately conserved genes. This approach was made possible thanks to the availability of the first isolate (27), and this report illustrates the fact that isolation of microorganisms is a prerequisite for efficient and reliable molecular and genetic study in microbiology. Indeed, the genome walking approach requires a large quantity of DNA, which can only be obtained from cultured bacteria.

The calculated GC content of rpoB was similar to those found for other T. whippelii genes. These GC contents vary from 46.93 to 48.46% for intergenic spacers (8, 16) to 50.6% for groEL (22), to 53.65 to 53.69% for the entire ribosomal operon (18), and to 57.38 to 57.78% for the 16S rDNA (D. Goldenberger and R. Lucchini, unpublished data; 18, 28) (GenBank accession no. AF202891). The last values in conjunction with phylogenetic analyses based on the 16S rDNA sequence led to a classification of T. whippelii within the high-GC content, gram-positive bacillus group.

In view of the frequency and potential severity of central nervous system involvement during the course of Whipple's disease, long-term treatment with rifampin has been advocated as a first-line antibiotic treatment for Whipple's disease (33, 37). Indeed, this antibiotic exhibits high penetration into the cerebrospinal fluid and cerebral tissue and is able to enter cells. However, routine experience (with a number of bacterial species) indicates that long-term therapy with rifampin alone results in the selection of rifampin-resistant bacteria. No rifampin susceptibility data for T. whippelii are available, and we cannot yet assess whether this rpoB sequence is that of a naturally rifampin-susceptible or naturally rifampin-resistant isolate. When rifampin susceptibility data for the T. whippelii isolate become available, interpretation of this sequence can be made in terms of genetic support for resistance. The development of an rpoB sequence-based test for the detection of genotypes encoding rifampin resistance directly from clinical specimens in the course of Whipple's disease may result from the present study.

Strictly speaking, phylogenetic inferences derived from a single-gene study cannot be extended beyond this particular gene, and direct extrapolation to bacterial phylogeny can be erroneous. This concern has been previously illustrated by the observation that phylogenies based on different genes show discrepancies (3), as in the case of the taxonomic relationships among Bartonella spp. (19, 29). This limitation can be addressed by applying high bootstrap values to every topology derived from this gene and by retaining topologies common to at least three different genes. Furthermore, base composition and codon usage differences among various bacterial strains constitute potential sources of inconsistencies (23, 34). Broad-spectrum phylogenetic studies based on comparison of highly conserved genes are subject to errors due to the GC content bias, the inherent ambiguity of nucleotide alignments, and the fact that reading frames are not taken into account in the alignment process. As a consequence, inferences based on comparisons of amino acid sequences of highly conserved proteins have been proposed to be more reliable than those based on the corresponding nucleotide sequences (13). We therefore analyzed rpoB-based taxonomic relationships of T. whippelii, thereby confirming the data previously obtained using ribosomal (16, 28, 39) and hsp65 (22) markers, although the paucity of the rpoB database limited the power of this analysis.

Alignment of the T. whippelii rpoB with those of closely related bacterial species allowed identification of rpoB regions specific to T. whippelii. We were therefore able to develop an rpoB-based molecular detection and identification method for this emerging pathogen for clinical material. When applied to the detection of T. whippelii DNA on clinical samples, this method achieved a 100% positive predictive value, although on a limited number of specimens (10 specimens). These results, however, proved that rpoB-based detection competes with the currently used 16S rDNA-based detection. Indeed, the fastidious nature of T. whippelii may prevent its isolation and culture from becoming a routine diagnostic tool in the immediate future. Furthermore, established PCR-based diagnostic methods will also be important for the effective evolution of newly described serological assays. Therefore, molecular detection of T. whippelii DNA is likely to continue to contribute to the diagnosis of the disease and to the selection of suitable clinical specimens for the isolation and propagation of additional strains. rpoB-based detection and identification of bacteria have emerged as an alternative to the 16S rDNA-based approach and have proven to be more discriminative than the 16S rDNA-based approach in some bacterial groups such as the enteric bacteria (20). Moreover, development of alternative molecular targets for the detection and identification of T. whippelii is necessary to circumvent the threat of molecular contamination when always relying on the same molecular target in the laboratory and to develop the "suicide PCR" protocol (26). In this respect, there is a considerable need for the development of numerous molecular-identification targets for the Whipple's disease-associated bacillus.