IS1111 insertion sequences of Coxiella burnetii: characterization and use for repetitive element PCR-based differentiation of Coxiella burnetii isolates

Coxiella burnetii is a gram-negative, obligate, intracellular bacterium that causes Q fever in humans. Infection occurs through inhalation of contaminated aerosols and fewer then ten organisms is sufficient to seed an infection [1]. Q fever can manifest as an acute infection characterized by a flu-like febrile illness, headache, and atypical pneumonia and the recommended treatment is doxycycline. A chronic infection is typified by endocarditis and/or hepatitis with poor prognosis [2].

Initial studies examining restriction fragment length polymorphisms (RFLP) in chromosomal and plasmid DNA from seven C. burnetii isolates revealed genetic heterogeneity among the isolates [3]. In subsequent research, RFLP patterns analyzed via sodium dodecyl sulfate-polyacrylamide [4] or pulsed field gel electrophoresis [5] allowed for the creation of six distinct genomic groups (I through VI). Isolates from acute disease patients could be placed in genomic groups I, II and III, while groups IV and V contained isolates from chronic disease patients. Furthermore, isolates placed in genomic groups I, II and III were shown to contain the plasmid QpH1, group IV isolates to contain the plasmid QpRS, and group V isolates are plasmidless [5], though QpRS-homologous sequences are located in chromosomal genome sequences of the group V isolates [6, 7]. The group VI isolates were initially reported to contain the plasmid QpDG, though newer experiments suggest these may actually possess QpH1, and QpDG containing isolates were shown to be avirulent in guinea pigs [8, 9]. Additional studies have characterized a variety of isolates using RFLP/pulsed field gel electrophoresis with differing restriction enzymes [10], multispacer sequence typing [11], multiple loci variable number of tandem repeats analysis [12, 13], and microarrays [14] which have discriminated isolates into as many as 36 distinct genotypes.

With the completion of the genome sequence of Nine Mile Phase I (9Mi/I), 20 unique copies of the IS1111 transposase were identified [15]. Previous work describing the insertion sequence noted that the coding region of the transposase is bounded by two sets of terminal inverted repeats at each end, which were previously referred to as the inner and outer inverted repeats [16]. The inner inverted repeats are thought to represent the termini of the IS element. It is predicted that IS1111 forms a circular intermediate, whereby the inner inverted repeats are brought closer together to form a strong promoter which increases transposase expression. The outer inverted repeats are believed to be chromosomal sequences which form a stem-loop structure that acts as a recognition site for insertion of the IS element but are not part of the IS element itself. This chromosomal target sequence is present more than 50 times throughout the 9Mi/I genome [17]. Sequence differences in the various copies of IS1111, most of which are present in the loop of the stem-loop recognition site, led to its further subdivision into IS1111 A, IS1111 B and IS1111 C [16]. We sought to compare the locations of IS1111 in a variety of isolates of C. burnetii to that of the 20 copies present in 9Mi/I using a PCR-based method. Because of the high copy number of the IS element in the C. burnetii genome and the unique regions adjacent to each element, this has proved to be a useful tool in differentiating isolates of C. burnetii into groups, much like the six genomic groups proposed previously [4, 5]. We present an algorithm which distinguishes members from five of the six genomic groups using four primer pairs and a two-step PCR cycling program.

PCR analysis included 14 isolates, 10 of which had previously been characterized as members of genetic groupings I through V [Nine Mile phase II (9Mi/II), Nine Mile RSA 514 (RSA 514), Ohio, Australian QD, Q195, Henzerling RSA 343 (Henz I), Henzerling, RSA 331 (Henz II), M44, KAV, PAV, Q238 and WAV], and 4 [Nine Mile Baca (9Mi/Baca), Scottish, Q238 and WAV] which had not been genotyped [4]. Our repository lacks the group VI Dugway isolates, thus these were not included. The analysis was completed for each isolate for each of the 20 IS elements present in 9Mi/I using the consensus primer IS1111-1 paired with a unique primer upstream of each of the IS elements. The presence or absence of certain IS elements was generally consistent among members of a particular genogroup, but allowed their differentiation from members of other genogroups. The results from each of these 14 isolates are provided (see Additional file 1) along with the element's orientation and nucleotide position in 9Mi/I [GenBank: AE016828], and GenBank accession numbers for the sequences determined in this study.

Genomic group II isolates

PCR products amplified using DNA from isolates Henz I, Henz II, and M44 were identical in size to 9Mi/I for 15 of the 20 IS elements (2, 3, 5, 6, 7, 8, 9, 11, 12, 13, 15, 16, 17, 18, and 19). IS elements 14 and 20 were confirmed absent in these isolates by sequencing, though our M44 isolate also possessed a subpopulation which contained IS 14 (Fig 1). In regards to IS 14, the deletion totaled 1,374 bp, which corresponds exactly to the predicted length of the IS element [17]. Also within this region, Henz I and Henz II, but not M44, contain a point mutation. The IS 20 deletion in these three isolates totaled 2,246 bp.

Amplicons to IS elements 1 and 4 were present but of larger size than 9Mi/I. In all three group II isolates, indels consisted of small deletions and slightly larger insertions. The IS 10 region, however, was quite unique in that PCR products using primer pairs IS 10 and IS1111-1 produced a product of incorrect size. Upon amplification with primer pairs 1269R and 1273R, which correspond to genes flanking the IS 10 element in 9Mi/I, products were obtained. Sequence data indicated that the site where the IS 10 primer anneals is missing due to a deletion of 1,650 bp in these isolates, but that the IS element is indeed present.

In this study, PCR amplification and DNA sequencing of the IS1111 transposase elements in the genome of C. burnetii and their surrounding sequence has been used to extensively characterize these elements in multiple isolates, and as target regions for inter- and intra-genomic analysis. In some isolates, IS element PCR resulted in products either larger or smaller than expected based on the 9Mi/I genome and these were generally due to small indels in the isolates tested. Numerous IS elements present in the 9Mi/I genome were absent in the isolates evaluated, mostly due to large deletions. Of the 25 deletions which could be completely characterized by sequencing, 16 were 1,374 in length. Larger deletions were detected at IS 20 (2,246 bp) in the group II isolates and at IS 7 (1,458 bp) in the group IV and V isolates, while a smaller deletion (1,341 bp) was detected at IS 9 in the group IV isolates. The 1,374-bp length corresponds exactly with the predicted length of the IS element [17], demonstrating that the IS element has likely failed to integrate into these sites thus far. The presence of larger deletions may indicate that the IS element inserted into certain sites and subsequently excised, taking additional genomic sequence with it or causing genomic rearrangements when doing so. Numerous nucleotide polymorphisms were also seen in the stem-loop portion of the IS insertion target sites (data not shown) which may render the sites unrecognizable by the transposase, or alternatively, the IS element simply may not have inserted into these IS recognition sites. The sequence data demonstrate that, where polymorphisms were noted, the stem of the stem-loop structure is most times preserved while changes are evident in the loop region. An understanding of target site recognition and insertion mechanisms will help to determine the importance of the nucleotides within the loop region. If it is found that the overall structure but not specific nucleotide sequence is key to recognition, the division of IS1111 into IS1111 A, IS1111 B and IS1111 C, as previously suggested based on sequence variation, may be unnecessary [16].

Analysis of the M44 isolate was complicated by evidence of a mixed population when analyzing IS 14. This result could indicate that the isolate is truly a mixed population, which is quite possible since it was never cloned. Previous work from our laboratory supports this hypothesis, as prior analysis of this isolate by an indirect immunofluorescence assay demonstrated epitopes reactive with both phase I- and phase II-specific antibodies [18]. However, this result could also occur if IS1111 is actively transposing in the genome, in which case circular intermediates should be detectable. PCR to determine if circular forms of IS1111 were present in 9Mi/I did not result in a product, though the linear form of IS1111 appeared to be expressed as judged by reverse transcriptase-PCR analysis (data not shown). It may be that expression of this linear form leads to an actively transposing IS element at low levels that we could not detect by PCR. The IS element appears rather stable, however, in that all 9Mi/I derived isolates have IS elements identically placed with no copies having been eliminated, though additional copies could be present and would not be detected by the methods used in this study. Certainly further research into the mechanism and frequency of IS1111 transposition is warranted.

The inter-genomic comparisons based on our IS element analysis showed very good correlation with RFLP analysis that was previously used to divide C. burnetii isolates into six distinct genomic groups [4]. Therefore, this analysis was used to develop a PCR-based genotyping assay which identifies an isolate as within one of the genomic groups, I through V. The assay was effective in discriminating genomic groupings in veterinary samples as well as purified isolates of C. burnetii. While most veterinary samples could be differentiated using the conditions initially set forth, an inhibitor was found to block amplification in three of the veterinary samples, as PCR products to all IS elements, even those conserved in all isolates tested, were not produced. Inhibition was easily circumvented by using a different master mix. It should be noted that inhibition occurred with both of the fecal samples that were tested. DNA from fecal samples is typically more difficult to amplify due to the presence of inhibitors and other factors, and if DNA from feces will be used for genotyping, extractions should be performed using protocols designed especially for feces [19]. Therefore, it is recommended that a control utilizing primers to a conserved IS element or a housekeeping gene also be included in the run to check for inhibition of the reaction. The DNA from any group I isolate should be used as a positive size control in applications of the algorithm, as group I isolates produce a product with all four algorithm primer pairs. Further, negative PCR results to any of the IS elements in the algorithm should also be verified by sequencing the region to ensure that mutations or deletions in primer binding regions are not at fault and that the IS element is truly lacking from this region. Sequencing of PCR products elucidated single nucleotide polymorphisms in IS element regions (data not shown) which could distinguish KAV from PAV, though no differences were observed between Henz I and Henz II or PAV and Q238. Sequencing was also effective at determining which IS elements were being amplified, and in some cases, sequences not found in the 9Mi/I genome were identified next to IS elements. It is interesting to note that in the group IV isolates, IS 10 and IS 18 primers annealed to sequences present in the Dugway 7E9-12 genome but not in Nine Mile. The origin of the Dugway strain and how the group IV isolates came to share genomic sequence with this strain, evolutionarily speaking, is intriguing. While other genomic changes undetectable by this method are surely present in these isolates, which would allow for further discrimination of isolates into a larger number of genotypes, the ability to quickly identify members of genomic groups I through V is useful. Perhaps a combination of rapid PCR-based methods such as the algorithm presented here, and VNTR (variable number tandem repeats) analysis [12, 13] or MLS (multispacer sequence) typing [11] for potentially higher resolution, will prove optimal.

It is interesting to note that the Qiyi isolate, typed as genomic group I in this study, was reportedly isolated from the bone marrow of a patient with chronic Q fever [20]. This represents a relatively novel finding, and supports a previous report that identified an isolate characterized as within genomic group I from a chronically infected human in France [10]. In contrast, most isolates from chronically infected humans have been classified into genomic groups IV and V. Clearly, further examination of the genomic composition and evolutionary relatedness of isolates from both acute and chronic human infections is warranted.

This study has begun to shed new light on the variations of IS elements within the genome of multiple isolates and the mechanism of IS element insertion in the C. burnetii genome. It also presents an algorithm which may provide a useful tool for grouping unknown isolates. Further, with the additional use of sequencing in these regions, a multitude of information can be learned about new isolates which may be used to determine their relatedness to known isolates of C. burnetii.