INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Coxiella burnetii, an obligate intracellular bacterium propagating in the phagolysosomes of eucaryotic cells, is the only species of the genus Coxiella. In comparison to other members of the family Rickettsiaceae, C. burnetii demonstrates high resistance to chemical and physical agents, making it possible for the organism to remain infectious after years outside the host cell (2). C. burnetii infection of humans manifests itself as acute Q fever with influenza-like symptoms. However, infection may also result in the therapy-resistant chronic form of Q fever with endocarditis, granulamatous hepatitis, or osteomyelitis (37).

Though C. burnetii has been recognized worldwide as an important pathogen, little is known about the genes that contribute to virulence, particularly tissue invasiveness and intracellular persistence. Virulence potential is correlated with lipopolysaccharide (LPS) content of the organism (55). When propagated in nonimmunocompetent systems such as embryonated chicken eggs or persistently infected cell cultures, C. burnetii converts from phase I to phase II particles. Phase shift is accompanied by a drastic change in LPS content and structure of the outer membrane, with a significant decrease in virulence potential (34). Initial studies of LPS phase variation focused on possible alteration of plasmid-encoded genes. However, cloning and sequencing of the entire QpH1 plasmid (45) and plasmid sequences integrated into the chromosome of plasmidless C. burnetii Scurry Q217 (51) revealed no evidence for genes involved in phase variation. Furthermore, plasmid type had been correlated to the type of disease (39), but this has been refuted by several authors (44, 58). These findings suggest virulence factors to be chromosomally encoded. Restriction fragment length polymorphism (RFLP) patterns demonstrated a great heterogeneity of C. burnetii isolates (21, 24, 46) which may be associated with virulence potential. A physical map of C. burnetii Nine Mile would provide a means to evaluate the reason(s) for this great heterogeneity and the different virulence potentials of isolates.

Due to restrictions imposed by the intracellular nature of C. burnetii, genetic studies have always been performed with Escherichia coli as the vehicle for gene expression. Nevertheless, gene organization of an organism can be studied only with an existing genetic map. Classical genetic maps are constructed by well-established methods such as transduction, transformation, conjugation, or transposon mutagenesis. But since obligate intracellular bacteria are not amenable to conventional linkage mapping, the only way to construct a genetic mapapart from sequencingis to first establish a physical map and then locate genes on the physical map by hybridization techniques. A first step toward generation of C. burnetii mutants was recently undertaken by Suhan et al. (42), with the successful transformation of C. burnetii to ampicillin resistance. Here we present the physical and genetic map of C. burnetii Nine Mile constructed mainly by two methods, pulsed-field gel electrophoresis (PFGE) and PCR.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains. C. burnetii Nine Mile RSA 493 was obtained from L. P. Mallavia, Washington State University, Pullman. C. burnetii NotI/Sau3a and NotI/EcoRI fragments were shotgun cloned in phagemid vector pBluescript II KS(+) (Stratagene, Heidelberg, Germany) and transformed into E. coli XL1 Blue (Stratagene).

PFGE sample preparation. Heat-inactivated (15 min, 85°C) C. burnetii organisms were diluted to a concentration of 2 × 10⁹ particles/ml. One volume of this suspension was mixed with 1 volume of solubilized 1% InCert agarose (Bio-Rad, Munich, Germany) at 50°C. Agarose-embedded C. burnetii was lysed overnight at 56°C with proteinase K (500 µg/ml) and washed twice with 10 volumes of 1× TE (10 mM Tris-HCl, 1 mM EDTA) for 30 min. Proteinase K was inactivated by phenylmethylsulfonyl fluoride (1 mM) at 50°C. C. burnetii total DNA embedded in agarose plugs was digested with 10 U of NotI/ml in 400 µl of 2× Universal buffer (Stratagene) overnight at 37°C.

PFGE. All PFGE gels (1% MBC agarose; Bio-Rad) were run on a CHEF (contour-clamped homogeneous electric field) mapper apparatus (Bio-Rad). The following parameters were applied to separate NotI fragments of the indicated size ranges: (i) 2 to 270 kb, pulse time of 0.1 to 11.0 s for 8 h, linear gradient and then 9.0 to 24.0 s for 12 h, linear gradient at constant voltage (6 V/cm) and constant angle (120°); (ii) 50 to 270 kb, pulse time of 14.23 to 16.08 s for 23 h 22 min, linear gradient at constant voltage (6 V/cm) and constant angle (120°), ramping factor of 1,357; and (iii) 2 to 55 kb (field inversion gel electrophoresis [FIGE]): pulse time of 0.05 to 0.12 s for 9 h 26 min, forward voltage of 9 V/cm, reverse voltage of 6 V/cm.

DNA probes. Biotinylated gene probes were prepared by PCR with biotin-21-dUTP (Amersham, Braunschweig, Germany), using different dTTP/biotin-21-dUTP ratios (3:1, 6:1, and 9:1). From cloned DNA fragments, plasmid DNA was prepared and biotin labeled by nick translation (Nick translation biotinylation kit; Serva, Heidelberg, Germany).

Southern hybridization. To dissect large NotI fragments, PFGE gels were exposed to UV light (Stratalinker; Stratagene) with an energy of 60 mJ/cm². Southern blotting was performed as downward blotting (8). Fragments were denatured (0.5 M NaOH-1.5 M NaCl, two cycles of 30 min each) and transferred to Biodyne B membranes (Pall; Dreieich, Germany) with 20× SSC (3 M NaCl, 0.3 M sodium citrate) as the transfer buffer. Transferred DNA was immobilized by exposure to UV light (120 mJ/cm²). Hybridization experiments were carried out at 60°C with 200 ng of denatured probe. Since probed DNA was always the same (C. burnetii Nine Mile) and only probes differed, the membrane was cut into stripes (3 by 130 mm) and hybridizations were performed in a specially constructed hybridization chamber. Stringency washes and chemiluminescence detection of biotinylated DNA were carried out as instructed by the manufacturer (Serva).

XL PCR. XL (extra-large) PCR was developed for the amplification of DNA fragments of up to 40 kb (7). XL PCR was performed on a Perkin-Elmer thermal cycler (model 9600; Perkin-Elmer/ABI, Weiterstadt, Germany) in a total volume of 50 µl consisting of 1× XL buffer, 1.1 mM magnesium acetate, 0.4 µM each primer, 200 µM each deoxynucleoside triphosphate, 1 U of XL polymerase, and 10⁴ to 10⁶ DNA templates. To amplify the 27.5- and 11-kb fragments, the following conditions were applied (values for the 11-kb fragment are in parentheses): 15 cycles consisting of denaturation at 94°C for 15 s, annealing and extension at 68°C for 15 min and 30 s (60°C, 7 min); 15 cycles consisting of denaturation at 94°C for 15 s, annealing and extension at 68°C for 15 min and 30 s, with an autoextension time of 15 s per cycle (60°C, 7 min plus 15 s per cycle). Prior to PCR, template DNA was allowed to completely denature at 94°C for 1 min. After PCR, an additional extension step at 72°C for 10 min was applied to maintain fully double-stranded DNA.

Sequencing. Nonradioactive sequencing reactions were performed with a PRISM Ready Reaction Dye-Deoxy Terminator Cycle Sequencing kit (Perkin Elmer/ABI) as recommended by the manufacturer.

Sequence analysis. DNA sequence analysis was performed with the DNASTAR software package (DNASTAR Inc., London, England). Primers were designed with the OLIGO software program (Medprobe, Oslo, Norway) and synthesized on a model 381A DNA synthesizer (Perkin-Elmer/ABI).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Mapping strategy. The physical map of C. burnetii Nine Mile has been constructed by a top-down approach in combination with PCR technology.

Physical map. PFGE has been shown to be the method of choice for constructing macrorestriction maps of bacterial genomes (10) provided that appropriate rare-cutting restriction enzymes are available. Criteria for the selection of restriction enzymes have been developed by McClelland et al. (31). Most obviously, in bacterial genomes with G+C contents above 45%, the tetranucleotide sequence CTAG is extremely rare. Similarly, trinucleotides CCG and CGG are rare in genomes with G+C contents of less than 45%. The G+C content of the C. burnetii genome has been determined to be 43 to 45% (36). Therefore, we selected restriction enzymes SfiI (GGCCNNNN'NGGCC), NotI (GC'GGCCGC), FseI (GGCCGG'CC), and SmaI (CCC'GGG). Finally we applied restriction enzyme NotI, which produces 29 fragments, 25 of which are distinguishable on ethidium bromide-stained CHEF-PFGE gels (Fig. 1B). For optimal resolution, portions of the molecular weight size range where DNA fragments accumulated were expanded. Resolution of NotI fragments around the 200-kb region was enhanced by extending the pulse times (Fig. 1A). The 30-kb region was expanded by FIGE (Fig. 1C). The average resolution of the macrorestriction map is 72.5 kb, with 2.1 kb being the smallest and 268 kb being the largest NotI fragment. Altogether, the size of the C. burnetii chromosome is 2,103 kb (Fig. 2).

View larger version (73K): [in this window] [in a new window]   FIG. 1.   CHEF-PFGE of NotI-digested total DNA. Different CHEF-PFGE parameters (see text) were applied to separate NotI fragments of 2 to 270 kb (B), 55 to 270 kb (A), and 6.7 to 55 kb (C). Lines indicate regions where resolution of NotI fragments has been increased. Panel A demonstrates fragments ranging in size from 210 to 268 kb to be clearly separated. Nevertheless, the two 227-kb NotI fragments are not distinguishable. Resolution enhancement is even more striking in panel C, where the 52.3-kb double fragment (B) divided into 53.8- and 50.8-kb NotI fragments. Panel C also demonstrates the 27.5- and 32.1-kb NotI fragments to be double fragments, as indicated by increased band intensities. In contrast, band intensity of the QpH1 plasmid suggests that there is only a single copy of the plasmid in C. burnetii Nine Mile.

View larger version (45K): [in this window] [in a new window]   FIG. 2.   Physical and genetic map of the chromosome of C. burnetii Nine Mile. NotI fragments are indicated by numbers on the outer circle (sizes in kilobases): 1, 7.3; 2, 50.8; 3, 13.8; 4, 26.1; 5, 227; 6, 227; 7, 103; 8, 27.5; 9, 151; 10, 16.9; 11, 268; 12, 9.4; 13, 3.9; 14, 210; 15, 19.2; 16, 6.7; 17, 168; 18, 6.7; 19, 28; 20, 53.8; 21, 12.4; 22, 27.5; 23, 32.1; 24, 118; 25, 5.3; 26, 32.1; 27, 8.2; 28, 2.1; 29, 240. NotI recognition sites are indicated by lines connecting the inner and outer circles. 1, contains the NotI recognition site. 2, hybridization signal intensity suggests IS1111a to exist at least once on the 50.8- and 53.8-kb NotI fragments (in Fig. 1B, lane 1, indicated as the 52-kb fragment). *, Could not be related to one of the 227-kb NotI fragments; hybridization signal intensity suggests IS1111a to exist at least once on both 227-kb NotI fragments (Fig. 1B, lane 1). **, putative genes located around the oriC region: fmu, glysAB, ygi2, ygi1, omp, gidB, gidA, 50K, 60K, and 9K.

Genetic map. ORFs with homology to database entries (putative genes) and genes were located on the physical map by Southern hybridization. Hybridization with the omp gene as a probe revealed a hybridization signal with the 27.5-kb NotI double fragment. To relate the omp gene to one of the 27.5-kb NotI fragments, XL PCR was performed with primers deduced from the omp gene and the ends of the 27.5-kb fragments. The positive PCR result with one primer combination was confirmed by hybridization and sequencing of the 11-kb amplicon. In total, we mapped 54 genes and putative genes, of which 39 (Table 2; Table 1, amplicons 17, 20, and 26) were extracted from the EMBL database and 16 putative genes (Table 1, amplicons 1, 3, 5, 8, 10, 11, 13, 14, 16, 18, 19, 23, 24, 25, 27, and 28) were newly identified during the mapping experiments. Nineteen putative genes with homology to database entries contained the NotI recognition sequence. Several putative polypeptides demonstrated only low homology to database entries (bcr, cmlA, and nlpD), though hydrophobicity plots were very similar. One gene, recognized as transposable element (23), was found to hybridize with at least nine NotI fragments.

View larger version (15K): [in this window] [in a new window]   FIG. 3.   Gene organization around the putative oriC of C. burnetii (B) compared to that of P. putida (A) and B. subtilis (C). The region upstream the oriC is almost identical to that of P. putida and B. subtilis (indicated by black boxes). Hypothetical proteins YGI1 and YGI2 of C. burnetii are homologous to YR55 and SpoJ of B. subtilis. Gene organization downstream the putative oriC of C. burnetii is quite different from that of P. putida and B. subtilis, as indicated by white boxes. Whereas in P. putida and B. subtilis genes around the oriC are transcribed in opposite direction, genes in C. burnetii are all transcribed in the same direction (indicated by arrows). Question marks above the white boxes indicate polypeptides without designation and with unknown function.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Since the introduction of PFGE (40), strategies for constructing physical maps changed from bottom-up (25, 57) to top-down techniques (10). With bottom-up techniques, genome maps are established by sorting clones from a genomic library. But this procedure is tedious and may become ambiguous if one enters regions of repetitive DNA. Moreover, regions which are difficult to clone represent one of the major problems encountered in bottom-up genome mapping. In contrast, top-down approaches are easily to perform and only few restriction fragments have to be examined for their natural order. Once the physical map is established, locations of a wide variety of other restriction enzymes can be placed on the existing map with relatively little effort.

The macrorestriction map of C. burnetii Nine Mile has been constructed mainly by two methods, PFGE and PCR. Southern hybridization was used to relate cloned fragments to the corresponding NotI fragments, and PCR was used to identify linking fragments.

The size of the C. burnetii Nine Mile chromosome has been determined from PFGE data to be 2,103 kb and thus has been underestimated by several authors (14, 36). Myers et al. (36) used renaturation techniques to determine the chromosome size, and Frazier et al. (14) applied PFGE. Both groups calculated a size of 1,600 kb for the C. burnetii chromosome, which is about 25% less than found in this study. Most obviously, Frazier et al. did not recognize double fragments and/or failed to notice fragments in regions where DNA fragments accumulated. Double fragments in this study were primarily identified by increased band intensities in PFGE gels compared to single bands of similar size. Duplication of these fragments was proven by hybridization to a PFGE blot of C. burnetii DNA digested with a second enzyme, by partial digestion, or by XL PCR. Other mapping strategies use two-dimensional PFGE to circumvent the need for detection of double fragments. However, one general intrinsic problem with two-dimensional PFGE is the reliable detection of smaller fragments due to unfavorably low fluorescence signals of the ethidium bromide stain. Smaller fragments may even diffuse out the agarose plug during digestion.

In most cases, genomes were mapped by two different techniques whereby most often partial digests were combined with the linking clone approach. In this study, linking clones were identified by two PCR methods (checkerboard and adaptor) instead of screening conventional DNA libraries or applying cross-hybridization techniques. The PCR methods circumvent some of the problems associated with cloned linking fragments. For instance, if two unlinked NotI-containing fragments are coligated, aberrant linking clones can be obtained. Cross-hybridization techniques use gel-purified probes to identify linking fragments. Nevertheless, gel-excised fragments contain impurities of randomly sheared DNA, leading to increased background hybridization signals. This increased background may hamper the unambiguous detection of faint hybridization signals generated from probes with very short overlaps. Moreover, in regions where DNA fragments accumulate, it is almost impossible to excise gel plugs without contamination with neighboring bands. Another restriction may be the amount of DNA in this fragment required to produce a reliable probe. A prerequisite to identify linking clones by PCR methods is the availability of sequences to construct primers. Sequences herein were obtained from shotgun cloning experiments. To make sure that clonable sizes are generated after digestion of C. burnetii total DNA, we chose as the second restriction enzyme (apart from EcoRI) Sau3A, which produces DNA fragments below 2 kb in size.

It remains uncertain whether the C. burnetii chromosome is linear or circular. However, several experimental results indicate the presence of a linear chromosome. Thus, it was not possible to clone the ends of the two fragments which would constitute the right and left ends of the linear chromosome, nor was it possible to amplify one of the putative ends (the 7.3-kb fragment) by XL adaptor PCR. In addition, undigested C. burnetii total DNA migrates as a single band on PFGE gels (data not shown), whereas open circular DNA molecules larger than 15 kb fail to migrate in PFGE (28). To date, linear chromosomes have been found in Borrelia burgdorferi (13), Streptomyces lividans (27), Agrobacterium tumefaciens (3), and Rhodococcus fascians (11), but nothing is known about the mechanism of replication in bacteria with linear chromosomes.

The oriC of C. burnetii has not yet been determined unequivocally. Gene organization around the oriC varies according to the taxonomic classification of the organism. Nevertheless, certain bacterial genes involved in replication are clustered. Whereas in all other organisms investigated so far, genes dnaA, dnaN, and gyrB are clustered and in the vicinity of rpmH, none of these genes is present downstream the oriC of C. burnetii (Fig. 3). This unusual gene organization reflects a replication mechanism quite different from that observed in bacteria with circular chromosomes. Therefore, further investigations are feasible to characterize the nature of the ends (e.g., single-stranded loops and telomeric structure) of the linear chromosome. The gyrA gene occupies a central position on the physical map of C. burnetii Nine Mile as has also been demonstrated for the gyrA genes of, for example, Borrelia burgdorferi, Clostridium perfringens, B. subtilis, and Leptospira interrogans (10). Nevertheless, presence of the dnaA gene, which is essential for replication and which is clustered with the gyrA gene in above-mentioned bacteria, remains to be confirmed.

The genetic map of C. burnetii revealed one gene which is recognized as a transposable element to be evenly distributed over the whole chromosome. Hoover et al. (23) demonstrated IS1111a to be present at least 19 times on the C. burnetii Nine Mile chromosome. We have mapped nine copies of the IS1111a gene, indicating that the gene appears more than once on several NotI fragments. Primers derived from the IS1111a element were used to establish a diagnostic PCR (52). Housekeeping genes belonging to the Krebs cycle show gene organizations identical to that of, for example, E. coli (12) or Azotobacter vinelandii (50).

On the C. burnetii chromosome there is only a single copy of the 16S rRNA gene, located on the 118-kb NotI fragment. This finding is in agreement with results published by Afseth and Mallavia (1) and correlates with the slow growth rate of C. burnetii. Though no direct evidence relates the number of rrn operons in an organism to growth rate regulation, slowly growing organisms such as Spiroplasma citri (56), Mycoplasma pneumoniae (26), and Chlamydia trachomatis (4) have been shown to have fewer copies.

As has been shown by DNA solution hybridization (29), C. burnetii isolates have many commonalities. However, RFLP analysis demonstrated a great heterogeneity in restriction patterns (21, 24, 46), possibly due to missense mutations and/or restriction site redistributions. Redistributions again are a result of significant chromosomal rearrangements (e.g., translocations, inversions, insertions, or deletions) which may play an important role in pathogenesis or virulence (47, 48). Similar rearrangements and exchange of DNA blocks have been described for Pseudomonas aeruginosa (38), leading to a 10% variation in chromosome size. Variation in chromosome size is even more striking in C. burnetii isolates ranging from about 1,500 to 2,400 kb. The high degree of DNA homology among C. burnetii isolates suggests that deletions resulting from recombinational (homologous or heterologous) events generated smaller genomes from larger ones. The now existing library of NotI linking fragments will facilitate comparisons of genome organization of different isolates and thus help to elucidate the reason(s) for the great heterogeneity in RFLP patterns. Furthermore, linking fragments may help to clarify which parts of the chromosome are the most conserved or variable ones and may help to answer the question of whether smaller C. burnetii genomes resulted from deletions in larger genomes. If so, this would provide a means for extensive epidemiological studies to evaluate the ubiquitous abundance of C. burnetii. The genetic map of C. burnetii Nine Mile may serve as a basis to identify genes or gene clusters affected by this tremendous loss of genetic information which seems to be no longer necessary for the organism to survive. It has been shown for Borrelia species that gene order is conserved, though nearly all isolates investigated demonstrated unique RFLP patterns (6). Whether this is also true for C. burnetii will be evaluated through comparison of gene order between isolates with different RFLP patterns.